Development and Optimal Probe Selection of an In Situ Penetration and Shear Apparatus for the Lunar Surface

Abstract

Precise in situ characterization of the mechanical properties of lunar regolith is critical for future lunar base construction and resource exploitation. However, existing detection methods predominantly rely on indirect inversion from rover wheel-soil interactions, which exhibit limitations in accuracy, real-time capability, and detection depth. Furthermore, specialized automated equipment capable of adapting to the complex lunar surface environment remains lacking. To address these challenges, this study presents the design and development of a novel autonomous in situ penetration-shear apparatus. The device automatically executes penetration and shear operations while recording real-time data, with a maximum penetration force of 25 N, shear torque of 2.5 N·m, penetration depth of 300 mm, and rotation angle of 360°. Given the maximum normal load constraint of 16 N imposed by the lunar rover platform, 24 probe configurations—varying in conicity, projected area, and vane number—were systematically evaluated using lunar soil simulants with three particle size distributions and two density levels. Multi-objective optimization was conducted to maximize detection efficiency, specifically penetration depth and shear torque, subject to a lightweight payload constraint (16 N). The multi-objective optimization reveals a fundamental trade-off: smaller conicity angles and projected areas favor deeper penetration, while larger projected areas enhance shear torque response. Under the 16 N constraint, the Pareto analysis identifies that a combination of moderate projected area, small conicity, and fewer vanes achieves the most balanced performance across all soil conditions. Results further demonstrate that increasing particle size and density substantially suppress both penetration capability and shear torque response, with compaction being the dominant factor limiting probe advancement under constrained normal loading. Results indicate that the optimal probe configuration comprises a 15° conicity, 324 mm² projected area, and two vanes, achieving an average penetration depth of 51.61 mm and average shear torque of 0.06 N·m across all test conditions. This study validates a complete automated system for characterizing lunar soil mechanical properties and provides an efficient, reliable hardware solution for future unmanned lunar exploration missions through optimized probe design. These findings establish a solid technical foundation for deep, high-precision in situ investigation of lunar soil structure and mechanical parameters, with significant implications for lunar base site selection and In Situ Resource Utilization (ISRU).

1. Introduction

Deep space exploration serves as a vital pathway for humanity to unravel the mysteries of the universe. Following the successful completion of the ‘orbit, land, and return’ three-phase strategic objectives of the China Lunar Exploration Program (CLEP) in 2020, the national deep space exploration roadmap has progressively shifted toward establishing a long-term lunar surface research station and preparing for crewed lunar missions [1,2]. In this new phase, the engineering geological environment of the lunar surface has emerged as a critical issue that must be addressed as a priority. Lunar regolith blankets the vast majority of the Moon’s surface, and its mechanical properties [3,4]—such as bearing capacity, shear strength, and relative density—exhibit a high degree of spatial variability across different regions. However, existing survey data remain insufficient [5], introducing significant uncertainties into the safe landing of future spacecraft, the trafficability of rovers, the development of in situ resource utilization (ISRU) systems, and the site selection and construction of scientific research stations [6]. Accordingly, the CLEP has elevated the refined, quantitative characterization of the physical and mechanical properties of lunar regolith to a top priority. For instance, one of the core objectives of the Chang’e-8 mission, scheduled for launch around 2028, is to conduct in situ mechanical measurements of the lunar polar environment using dedicated payloads—marking a clear shift in which engineering demands are driving scientific exploration toward increasingly targeted objectives.

At present, the measurement of mechanical properties of loose surficial media on planetary bodies primarily encompasses three technical approaches: offline laboratory testing, indirect inversion based on moving components, and direct measurement using dedicated in situ devices. Laboratory methods such as triaxial tests and direct shear tests are well-established techniques in terrestrial geotechnical engineering for obtaining soil strength and deformation parameters. These methods can systematically characterize the mechanical response of soil specimens under controlled stress paths and hold significant value in the study of constitutive parameters. However, such methods typically rely on sampling, encapsulation, sample return, or in-cabin preprocessing, are sensitive to sample disturbance, and involve bulky and heavy equipment with complex clamping and confining pressure loading systems, making them difficult to adapt directly to unmanned lunar surface missions that demand low-gravity compatibility, high-vacuum tolerance, and a high degree of automation. More critically, laboratory tests reflect the properties of reconstituted or disturbed specimens obtained “post-sampling,” and cannot fully preserve the in situ stratification, pore structure, and inter-particle contact characteristics of the lunar regolith [7,8]. Consequently, they are inherently limited in their ability to characterize the true in situ conditions of the lunar surface.

Another commonly adopted approach relies on existing contact components of rovers and landers—such as rover wheels [9,10], lander footpads, or sampling mechanisms [11]—to indirectly invert lunar regolith parameters by analyzing sinkage, slip ratio, traction force, and wheel track morphology in conjunction with terramechanics models [12,13,14]. For example, NASA’s Spirit, Opportunity, and Curiosity Mars rovers derived mechanical parameters by analyzing wheel–soil interaction data [9], while the Viking and Phoenix landers estimated soil properties using forces exerted by their sampling arms and footpads [11]. It should be noted that lunar regolith differs fundamentally from Martian soil due to the unique space weathering environment on the airless Moon, where solar wind implantation and micrometeorite bombardment continuously modify grain properties, producing agglutinates and nanophase iron that alter the mechanical behavior. Recent studies have shown that regolith maturity and physical state exhibit pronounced spatial variability across the lunar surface, as evidenced by correlations between boulder distribution, circular polarization ratio, and optical maturity [15], further emphasizing the need for direct in situ mechanical characterization at specific sites of interest. China’s Yutu rover on the Chang’e-3 mission and Yutu-2 rover on the Chang’e-4 mission employed similar approaches, extracting valuable mechanical and morphological data on lunar regolith through analysis of slip and sinkage measurements and high-resolution imagery [3]. Nevertheless, the fundamental limitation of indirect inversion methods lies in their strong dependence on model assumptions and boundary conditions. The outcomes are susceptible to the coupled effects of wheel lug geometry, load distribution, slip conditions, local terrain variability, and particle-scale heterogeneity, which collectively give rise to pronounced non-uniqueness in parameter inversion. Furthermore, the detection depth of such methods is generally confined to the shallow wheel–soil contact zone, making it difficult to actively acquire continuous mechanical responses over a specified depth range. These methods also lack dedicated loading paths specifically designed and optimized for the measurement objectives of the exploration mission itself.

Therefore, compared with indirect estimation methods that rely on existing moving components, dedicated in situ measurement devices are better suited to undertake refined mechanical surveying tasks on the future lunar surface. Such devices are capable of acquiring high-quality data, with notable examples including the Self-Recording Penetrometer (SRP) deployed during NASA’s Apollo missions [16,17,18,19] and the cone–vane integrated tester (PROP) carried aboard the Soviet Lunokhod rovers [20]. In contrast to laboratory equipment such as triaxial and direct shear apparatuses [21], in situ devices eliminate the need for complex sampling procedures and confining pressure systems, enabling mechanical measurements while preserving the natural packing state of the lunar regolith. This makes them considerably more compatible with the requirements of future unmanned lunar missions in terms of lightweight design, low power consumption, and automation.

However, existing in situ lunar regolith measurement devices still have room for further development. First, historical lunar-surface-dedicated instruments were largely constrained by the mission conditions and technological capabilities of their era. They fell short of meeting the modern demands for high precision, continuous data acquisition, and intelligent operation in contemporary lunar exploration, particularly in terms of dimensions, degree of automation, data sampling capability, and repeatability. Second, future lunar rover platforms typically impose stringent constraints on payload mass and normal loading capacity. In particular, under low-gravity conditions, the effective normal load that can be applied to the ground is relatively small, which means the conventional approach of “enlarging the probe to enhance the signal” is not necessarily applicable. Previous studies on soil shearing have demonstrated that probe dimensions significantly influence the soil failure mode and torque response, and that test results exhibit pronounced scale effects [22,23]. For lightweight lunar platforms, an oversized probe may reach the loading limit before achieving the required working depth. In contrast, an undersized probe, while advantageous for penetration, may yield weak shear signals and diminish the capacity for parameter identification. Consequently, achieving a balance between penetration capability and shear response by selecting optimal probe geometric parameters [24] under the constraint of limited normal loading [25] constitutes a key scientific and engineering challenge in the design of in situ mechanical measurement devices for the lunar surface [26]. In addition to penetration–shear testing, the upcoming CP-22 mission also aims to conduct drilling operations on the lunar surface, further reflecting the growing international emphasis on subsurface mechanical characterization and resource prospecting as essential precursors to sustained lunar presence.

From the perspective of engineering applicability, an integrated penetration–shear apparatus offers more pronounced comprehensive advantages compared with standalone penetration tests or standalone shear tests. The penetration process captures the resistance characteristics of lunar regolith under normal loading and provides a basis for identifying stratification features at varying depths [7,27,28]. In contrast, the shearing process further characterizes the local shear resistance response and furnishes supplementary information for inverting regolith strength parameters. The combination of both not only enhances the measurement dimensionality but also establishes a more complete chain of mechanical responses at a single test point, thereby strengthening the overall understanding of the bearing–deformation–failure behavior of lunar regolith. Moreover, the integrated apparatus can be readily incorporated into unmanned lunar rover platforms and, through programmable control, can achieve automated penetration, constant-angular-velocity shearing, real-time data acquisition, and telemetry to ground stations. It thus offers the merits of standardized testing procedures, good repeatability, and minimal manual intervention. These attributes are difficult to achieve simultaneously with conventional triaxial apparatuses, direct shear devices, or wheel-mechanics-based inversion methods.

Based on the above considerations, this paper addresses the practical requirements of “lightweight platforms, low normal loading, and the need for direct in situ measurement” in future unmanned lunar surface missions. It presents the design and development of an autonomous in situ penetration–shear apparatus for the lunar surface. The apparatus is capable of automatically executing probe penetration and rotational shearing processes while recording key data—including penetration force, shear torque, penetration depth, and rotation angle—in real time. Compared with triaxial and direct shear equipment, the apparatus offers the advantages of a more compact structure, less sample disturbance, no need for complex confining pressure or sampling procedures, and the ability to perform tests directly under in situ conditions. Compared with wheel-mechanics-based inversion methods, it provides well-defined loading paths, controllable test locations, directly measured response parameters, and lower interpretive uncertainty. Compared with historical single-function penetration or shear devices, the apparatus further achieves the integration of penetration and shear capabilities, automated control, and an interchangeable probe design, making it better suited to the operational scenarios of modern unmanned lunar missions. The apparatus further achieves the integration of penetration and shear capabilities—whereby a single probe is first advanced vertically into the regolith under controlled axial loading to measure penetration resistance as a function of depth, and subsequently rotated in place at the achieved depth to measure torsional shear resistance—enabling both measurements to be performed sequentially at the same test point through automated control of two independent motor-driven axes, along with an interchangeable probe design, making it better suited to the operational scenarios of modern unmanned lunar missions.

Building on this foundation, and considering the limited maximum normal load available from future lunar rover platforms, this paper further conducts a systematic experimental investigation on probe conicity, projected area, and the number of vanes (thin plate-like blades extending radially from the probe shaft that engage with the surrounding soil during rotation to mobilize shear resistance). A comparative analysis of 24 probe configurations is performed under simulated lunar regolith conditions encompassing three particle size ranges and two relative density states. By establishing a multi-objective evaluation model with maximum penetration depth and maximum shear torque as the core performance indicators, the optimal probe parameter combination suited to low-load operating conditions is subsequently identified. The objective of this study extends beyond the development of an in situ penetration–shear testing system for the lunar surface; it further seeks to determine which probe geometric parameters can achieve the best balance between “sufficient penetration” and “effective shear response” under constrained normal loading conditions. The findings are intended to provide a reference for the design of in situ regolith mechanical surveying payloads in the forthcoming Chang’e-8 mission and subsequent lunar exploration missions, and to offer fundamental technical support for the inversion of engineering parameters of lunar regolith and site selection for lunar surface construction.

2. Development of the Penetration Shear Apparatus

2.1. Hardware and Software Design

2.1.1. Penetration and Shear Unit

The conceptual design of the PC-based penetration–shear apparatus is illustrated in Figure 1. The apparatus comprises four main components: (1) a drive system that advances the probe into the simulated lunar regolith, (2) a drive system that rotates the probe at a prescribed angular velocity, (3) a sensor unit that measures the axial force and torque acting on the probe, and (4) a data recording system. The drive system of the penetration–shear apparatus consists of two DC brushed motors. The apparatus is mounted on a purpose-built test bench, with an overall length of approximately 650 mm.

For the penetration function, a DC brushed motor and leadscrew mechanism were employed to achieve vertical motion. The leadscrew is 350 mm long, allowing a penetration depth of 300 mm. To ensure the designed system can operate over its maximum linear displacement, the penetration sleeve is positioned at the very top in the initial state, with a proximity switch detecting whether it is at the upper limit. The DC brushed motor increases torque via a worm gear reducer, with the worm gear fixed to the leadscrew via a coupling. The encoder count for one rotation of the DC worm gear brushed motor output shaft is $M_1$, the real-time encoder count of the DC worm gear brushed motor is $M_{\mathrm{w}}$, the reduction ratio of the worm gear reducer is $A_1$, and the pitch of the leadscrew is $d$. With a pitch of 1 mm, for every single rotation of the worm gear output shaft, the T-shaped flange on the leadscrew moves 1 mm. If the angular displacement of the worm gear is ${\theta }_1$, the relationship between the change in the DC brushed motor encoder count and the movement distance $D$ can be derived using the formula.

$${\theta }_1={\displaystyle \frac{M_{\mathrm{w}}}{A\times M_1}}\times 360°$$

(1)

$$D=({\displaystyle \frac{{\theta }_1}{360°}})\times d$$

(2)

The rated speed of the DC brushed motor is 470 RPM. Based on the conversion using the aforementioned formula, the rated penetration speed can reach 47 mm/s.

For the shear function of the system, a DC brushed motor is installed at the lower end of the sleeve. This DC brushed motor increases torque via a planetary gear reducer (a compact gear mechanism in which multiple planet gears rotate around a central sun gear within a ring gear, achieving high reduction ratios in a coaxial, space-efficient configuration). The output shaft of the planetary gear reducer is connected to a Force-Torque Sensor using set screws, and the lower end of the Force-Torque Sensor is threaded to the probe. The encoder count for one rotation of the DC planetary brushed motor output shaft is $M_2$, the real-time encoder count of the DC planetary brushed motor is $M_{\mathrm{p}}$, the planetary gear reducer has a reduction ratio of $A_2$, and the angular displacement is ${\theta }_2$.

$${\theta }_2={\displaystyle \frac{M_{\mathrm{p}}}{A_2\times M_2}}\times 360°$$

(3)

The rated speed of the DC brushed motor is 11 RPM. Based on the conversion using the aforementioned formula, enabling a rated shear velocity of up to 66 degrees/s.

2.1.2. Penetration Shear Apparatus Processor Unit

The electrical equipment employed is listed in

Table 1. Specific Models of Electrical Equipment.

Components	Specifications
USB to RS485 Converter	Baud Rate Range: 1200 bps–6 Mbps RS485 Ports: four channels Power Supply: 5 V via USB
Force-Torque Sensor	Tension/Compression Force: 25 N Torque: 2.5 N·m
Multi-Channel Force Sensor Transmitter	Voltage Range: 12–30 V Control Signals: RS232, RS485 Channels: three channels
Aisikong (AQMD) DC Brushed Motor Driver	Model: AQMD6008NS-TBE Voltage Range: 9–60 V Control Signals: Potentiometer, Analog, PWM, Pulse, Frequency, Switch, RS485/CAN Applicable Motor: DC Brushed Motor
Worm Gear DC Brushed Motor	Rated Speed: 470 RPM Voltage: 24 V
Planetary Gear DC Brushed Motor	Rated Speed: 11 RPM Voltage: 24 V
Modbus RTU Remote I/O Analog Acquisition Module	Output Type: Relay Control Signal: RS485 Configuration: 4 Inputs/4 Outputs
Cylindrical Proximity Sensor	Diameter: 4 mm Output Type: NPN Normally Closed (NC)
Switching Power Supply	Output Voltage: 24 V Output Current: 4.5 A

. A penetration shear apparatus system was developed to read and store signals received from DC brushed motors and Force-Torque Sensors. The developed system consists of various components, including a USB-to-four-port RS485 hub, Aisikong DC brushed motor drivers, RS485 relays, and NPN normally closed proximity switches. The USB-to-RS485 hub communicates with the Aisikong DC brushed motor driver, the Modbus RTU protocol remote I/O analog acquisition module, and the force sensor multi-channel transmitter, respectively, via the A and B lines of RS485. The electrical schematic of the developed system is shown in Figure 2.

The Force-Torque Sensor is connected to the output shaft of the planetary-geared DC brushed motor via a set screw, while the other side is threaded to the probe. This setup is used to measure the received tension/compression force and torque. The measurement range for force is ±25 N, and the torque measurement range is 2.5 N·m. The power supply voltage for the Force-Torque Sensor is 5 V, which is provided by the force sensor multi-channel transmitter. The force sensor multi-channel transmitter operates on a supply voltage of 12–30 V. Its communication data acquisition speed can reach 320 times/second, with an internal resolution of 20 bits and a signal input range of ±20 mV. The Force-Torque Sensor directly acquires the physical quantities of force and torque, converting them into weak electrical signals. The force sensor multi-channel transmitter then standardizes and converts these signals from the sensor, unifying the output for data transmission via RS485 signals.

A worm gear DC brushed motor provides the driving power for the penetration mechanism. It features an integrated encoder used to acquire the current rotational angle. The motor interface consists of six pins: Motor Power (+), Motor Power (−), Sensor Ground, Sensor Power, Signal Output A, and Signal Output B. These pins are connected to the corresponding terminals of the Aisikong (AQMD) DC brushed motor driver. The driver receives control commands via RS485 communication. By parsing Modbus RTU instructions, the driver modulates the electrical output to control the worm gear DC brushed motor, enabling movement according to predetermined parameters for direction, speed, and distance (Figure 3).

The planetary gear DC brushed motor functions similarly to the worm gear motor described previously, with the primary distinction being the type of reduction gear used. This motor is designated for the probe’s shearing function and is mounted beneath the sleeve of the penetration shear apparatus. The planetary gear reducer is selected because its output shaft is coaxial with the input shaft, significantly reducing the overall radial dimensions of the motor assembly and making it suitable for confined spaces. Like the worm gear motor, the planetary gear DC brushed motor is controlled by an Aisikong (AQMD) DC brushed motor driver to execute its movement functions.

The system also incorporates a cylindrical proximity sensor of the NPN Normally Closed (NC) type. It features three pins: Sensor Ground, Sensor Power, and Signal Output. These pins are connected to a Modbus RTU remote I/O analog acquisition module (configured for digital input). The sensor detects the proximity or separation status of the T-shaped flange and outputs a corresponding digital signal. The remote I/O module captures this digital signal and transmits it to the host PC via RS485 communication.

The current ground prototype employs RS485-based communication for laboratory operation. For future lunar surface deployment, the control architecture will be adapted to support autonomous operation sequences compatible with the rover’s onboard computing and telemetry systems, enabling the apparatus to perform pre-programmed penetration–shear measurement cycles without requiring real-time ground-station intervention.

The sensor unit of the penetration–shear apparatus and its dedicated embedded system are shown in Figure 4. The sensor consists of an industrial cylindrical Force-Torque Sensor mounted between the output shaft of the shear motor at the lower end of the apparatus and the threaded end of the probe. The probe is interchangeable with different sizes and is connected via an M8 thread, enabling the measurement of penetration resistance and torque. The penetration depth of the probe is obtained by converting the encoder counts of the penetration motor, and the shear angle of the probe is derived from the encoder counts of the shear motor.

According to the previous equation calculation, the linear displacement accuracy in the penetration function is about 0.0018 mm, and the angular displacement accuracy in the shear function is 0.0096°. The Force-Torque Sensor has a force accuracy of 0.01 N and a torque accuracy of 0.001 N·m. The overall sampling frequency is approximately 20 Hz.

2.1.3. Development of PC-Side Host Software

To enable data recording during testing, a Python-based (Python 3.8) PC host software application was developed to control the motors and sensors of the penetration–shear apparatus and store measurement results on the PC hard drive. The graphical user interface was developed based on the open-source PySide6 framework. The user interface was first created using Qt Designer(Qt 6.5, PySide6), and the saved UI files were then converted into Python files using the pyside6-uic command. The resulting Python files were imported into the main program as window layouts. The application code was written in Python and packaged as a standalone executable (.exe) for deployment.

The operation interface is shown in Figure 5, which comprises four functional zones: Zone A for communication control, Zone B for data acquisition, Zone C for precise numerical data display, and Zone D for data visualization. The user interface of the developed PC host software is capable of displaying and storing force (N), torque (N·m), linear displacement (mm), angular displacement (°), time (ms), and date data. The data are stored on the PC in .xlsx format and can be subsequently processed using scientific computing software such as MATLAB (R2025a, MathWorks) and PyCharm (2025.1, JetBrains).

2.2. Calibration Setup and Procedures

2.2.1. Load Cell Calibration in Compression Mode

The calibration of the Force-Torque Sensor used for measuring the axial force was conducted in compression mode using an appropriate test setup. For the calibration of the force sensor, weights of varying masses were applied. The Force-Torque Sensor was first vertically fixed on a level tabletop, and its reading was then zeroed. Thereafter, weights were incrementally applied from 0 to 2.5 kg in steps of 0.5 kg. The calibration curve plotted between the force values acquired by the PC from the Force-Torque Sensor and the corresponding weights is shown in Figure 6. The results indicate that the force output of the Force-Torque Sensor in compression mode varies linearly with the count values registered on the digital force gauge, with a coefficient of determination (R²) of 0.99.

2.2.2. Torque Element Calibration in Shear Mode

The calibration of the Force-Torque Sensor used for measuring torque was conducted in rotational shear mode using an appropriate test setup. For the calibration of the torque sensor, a certified electronic digital torque wrench was employed. The Force-Torque Sensor was first connected to the digital torque wrench via a 3D-printed adapter, then placed horizontally on the ground and secured in position. The readings of both the Force-Torque Sensor and the digital torque wrench were then zeroed. Thereafter, torque was incrementally applied using the digital torque wrench from 0 to 2.5 N·m in steps of 0.5 N·m. The calibration curve plotted between the torque values acquired by the PC from the Force-Torque Sensor and the torque readings observed on the digital torque wrench is shown in Figure 7. The results indicate that the torque output of the Force-Torque Sensor varies linearly with the values registered on the digital torque wrench, with a coefficient of determination (R²) of 0.99.

2.2.3. Calibration of the Penetration Motor for Depth Measurement

The calibration of the penetration motor was performed on a leveled, simulated lunar regolith surface by slowly advancing the probe of the penetration–shear apparatus into the regolith. The penetration depth was increased incrementally from 0 to 270 mm in 30 mm increments. The calibration curve, plotted between the penetration depth observed in the PC host software and the depth manually measured with a metal ruler, is shown in Figure 8. As shown in Figure 8, an excellent correlation (R² = 1) exists between the penetration depth recorded by the PC host software and the manually measured depth.

2.2.4. Calibration of the Shear Motor for Conicity Measurement

The calibration of the shear motor was conducted under no-load conditions. The angular displacement was monitored on the developed PC host software. A reference mark indicating the origin was made on the surface of a rod adjacent to the probe. The probe was rotated incrementally from 0° to 180° in 30 ° increments. The calibration curve, plotted between the angular displacement observed in the PC host software and the angular displacement measured manually with a protractor, is shown in Figure 9. As shown in Figure 9, an excellent correlation (R² = 1) exists between the angular displacement recorded by the PC host software and the manually measured angle.

3. On-Site Experimental Optimization of Cone Tip Dimensions

3.1. Preparation of Simulated Lunar Soil

The experiments were conducted at the Key Laboratory of Bionic Engineering, Ministry of Education, Jilin University. The reference sample was the Apollo 16 return sample (highland region), whose mineralogical composition—dominated by anorthositic plagioclase—was used as the target for formulating the JLU-M series simulants. The simulants were produced from terrestrial source materials selected to approximate the mineral assemblage and chemical composition of highland lunar regolith and were subsequently sieved into three controlled particle-size ranges for systematic parametric testing.

The experiments utilized three lunar regolith simulants with different particle sizes—JLU-M-1, JLU-M-2, and JLU-M-3—developed at the Key Laboratory of Bionic Engineering, Ministry of Education, Jilin University (Changchun, China). The particle size ranges were selected based on the known granulometric characteristics of lunar regolith documented from Apollo return samples [29,30]. JLU-M-1 (<1 mm) represents the dominant particle size fraction of typical lunar regolith; JLU-M-2 (1–2 mm) corresponds to coarser fractions present in less mature regolith regions; and JLU-M-3 (2–4 mm) simulates extreme operating conditions involving coarse fragments, providing a conservative upper-bound test scenario for probe performance evaluation. The particle size distributions of the three simulants are given in

Table 2. Particle Size Distribution.

Simulant	JLU-M-1	JLU-M-2	JLU-M-3
Particle Size/mm	<1	1~2	2~4

. The simulants are shown in Figure 10, where it can be observed that the simulant with particle sizes less than 1 mm appears gray. Meanwhile, the simulant with particle sizes larger than 1 mm exhibits a black-and-white speckled appearance due to the larger grain size. The coarser simulant is intended to represent extreme operating conditions.

The angle of repose and the angle of collapse for the three types of simulated lunar soil were measured using a BT-1001 Intelligent Powder Characteristics Tester (Dandong Bettersize Instruments Ltd., Dandong, China). This tester, illustrated in Figure 11, consists of several primary modules: a feed funnel, a light source, a sample platform for the angle of repose/collapse, a platform base, a camera, and a square collection tray.

Three replicate tests were performed for each lunar regolith simulant, and the mean values of the angle of repose and the angle of collapse were calculated. The testing procedure for the angle of repose is illustrated in Figure 12. It consists of the following steps: (a) free accumulation stage, in which the sample is discharged through the outlet and gradually forms a cone on the angle-of-repose/angle-of-collapse sample platform; (b) static piling stage, in which the feeding is stopped once the sample has filled the platform to form a symmetrical cone with material falling on both sides of the platform axis; (c) tapping stage I; (d) tapping stage II; and (e) tapping stage III, during which an electromagnet initiates vibration and a camera automatically captures images to compute the angle of repose and the angle of collapse for each run. The final calculated results are summarized in

Table 3. Test Results of Conicity of Repose and Conicity of Collapse.

Simulant Type	No.	Conicity of Repose (°)	Conicity of Collapse (°)	Mean Conicity of Repose (°)	Mean Conicity of Collapse (°)
Simulant I	1	39.16	23.83	42.35	26.32
	2	43.73	26.11
	3	44.16	29.01
Simulant II	1	39.02	21.87	39.26	22.23
	2	39.23	22.84
	3	39.53	21.98
Simulant III	1	38.71	20.82	37.9	22.78
	2	36.63	23.9
	3	38.36	23.62

: JLU-M-1 has an angle of repose of 42.35° and an angle of collapse of 26.32°; JLU-M-2 has an angle of repose of 39.26° and an angle of collapse of 22.23°; and JLU-M-3 has an angle of repose of 37.9° and an angle of collapse of 22.78°.

The three lunar regolith simulants were loaded into separate acrylic cylinders and manually prepared into two states: dense and loose. The specific procedures are as follows. For the dense state, the acrylic cylinder was vibrated using a tapping machine for 15 min, after which the surface was leveled with a scraper to ensure uniformity, thereby producing a high-density simulant. In the loose state, the simulant was deposited naturally and then leveled with a scraper, resulting in a low-density simulant. The densities of the three simulants were determined from their volumes and masses, and are presented in

Table 4. Densities of the Three Types of Simulated Lunar Soil.

Density/(g/cm3)	Loose State	Dense State
Simulant I	1.37	1.71
Simulant II	1.27	1.37
Simulant III	1.44	1.51

.

3.2. Cone Tip Dimension Design

The geometric parameters of the cone tip significantly influence penetration resistance, local stress distribution, and the degree of test-induced disturbance. Accordingly, this paper selects the conicity and the vane diameter as the primary design variables for the cone tip geometry; the specific parameters are listed in

Table 5. Dimensions of the Cone Tip.

Name	Conicity (°)	Vane Diameter (mm)	Projected Area (mm2)	Radius (mm)
Cone Tip I	15	12.83	129	6.415
Cone Tip I	30	12.83	129	6.415
Cone Tip I	45	12.83	129	6.415
Cone Tip II	15	20.27	324	10.135
Cone Tip II	30	20.27	324	10.135
Cone Tip II	45	20.27	324	10.135
Cone Tip III	15	30	707	15
Cone Tip III	30	30	707	15
Cone Tip III	45	30	707	15
Cone Tip IV	15	35.7	1000	17.85
Cone Tip IV	30	35.7	1000	17.85
Cone Tip IV	45	35.7	1000	17.85

, and a schematic of the cone tip configurations is presented in Figure 13. In previous studies on static cone penetration of lunar regolith, cone tips with 30° and 60° conicities have already been employed [31]. In this study, conicities of 15°, 30°, and 45° are selected. Among these, 30° and 45° are representative conicities commonly used in penetration-type cone tips [32]. Meanwhile, 15° is included as a small-conicity extension to investigate the effect of an acutely tapered cone geometry on penetration capability and response sensitivity under low normal loading conditions. The base dimensions are designed with reference to the geometric scale of standard static cone penetrometer tips, while also taking into account the miniaturization requirements and structural space constraints of lunar in situ exploration devices. Four diameter levels are set at 12.83, 20.27, 30.00, and 35.70 mm, corresponding to radii of 6.415, 10.135, 15.000, and 17.850 mm, respectively. This results in a total of 12 cone tip configurations combining three conicitys with four diameters, which are used to systematically analyze the effects of cone tip geometric parameters on the penetration and shear response characteristics of lunar regolith.

The vane design is based on the number of vanes, vane shape, and vane dimensions. Two levels of vane quantity are considered: two and four vanes. A triangular vane shape is selected, with the vanes symmetrically distributed, yielding a total of two vane configurations (Figure 14).

3.3. Field Testing

The maximum allowable axial force was set at 16 N, corresponding to the normal load constraint imposed by the lightweight lunar rover platform planned for the Chang’e-8 mission. This rover, with a total mass of approximately 35 kg, provides a limited reaction load capacity at the rear-mounted payload interface. The 16 N threshold was determined by the mission engineering team through a comprehensive stability and trafficability analysis, ensuring that the penetration reaction force does not compromise the rover’s locomotion performance or tip-over safety margin on the lunar surface. Tests were conducted on the aforementioned different lunar regolith simulants. At each sampling point, the penetration–shear apparatus was first mounted vertically above the simulant surface to ensure that the probe was positioned close to the target. Teng et al. demonstrated that variations in loading conditions can significantly alter the stress–strain response, volumetric behavior, and apparent shear strength of regolith simulants, highlighting the importance of carefully controlled and reported test parameters [33,34]. A penetration velocity of 1 mm/s was prescribed, and the probe was advanced downward with a maximum allowable axial force of 16 N, corresponding to the lightweight payload constraint of a lunar rover. That is, the penetration motion was terminated once the force acting on the probe reached 16 N. This slow and constant penetration velocity is beneficial for improving the quality of force measurements. Subsequently, the shearing phase was initiated, during which the tip was rotated clockwise at an angular velocity of 3°/s through 180°. If the torque experienced by the probe exceeded 2.5 N·m during the angular displacement, the shearing motion was stopped. After the completion of each penetration and shearing test, the simulant was re-prepared to ensure that successive tests did not interfere with one another. Each test configuration was repeated three times, and the obtained maximum penetration depths and torques were averaged for subsequent analysis.

The cone tips of the penetration–shear apparatus were designed in multiple combinations to accommodate a range of testing conditions. Specifically, the design parameters of the cone tip include three conicities (15°, 30°, and 45°), four cone tip radii (6.415 mm, 10.135 mm, 15 mm, and 17.85 mm), and two vane quantities (two and four vanes), resulting in a total of 24 cone tip combinations. The test subjects comprised three lunar regolith simulants of different particle sizes, each prepared in both loose and dense states, thereby comprehensively capturing the influence of different simulant conditions on the test results. A full factorial experimental design was adopted in this study, in which all levels of each factor were fully crossed with all levels of every other factor, ensuring the systematicness and comprehensiveness of the experimental design. Each test configuration was repeated three times to enhance data reliability and result consistency, and all combinations were tested and analyzed individually. This design enables a thorough investigation of interactions among cone tip parameters, simulant particle sizes, and density states, and their effects on test results, thereby providing comprehensive and reliable experimental data to support further research.

3.4. Multi-Objective Optimization Model and Pareto Frontier Solution

The multi-objective optimization model aims to analyze experimental data to investigate the influence of structural parameters (conicity, projected area, and vane count) on performance indicators (maximum penetration depth and maximum torque) across different types of simulated lunar soil. The ultimate goal is to identify parameter combinations that demonstrate superior performance under a variety of simulated lunar soil conditions. The core objective of the model is to maximize penetration depth and torque while simultaneously minimizing their volatility (i.e., standard deviation) across different soil types, thereby ensuring that the recommended parameter combinations possess both high comprehensive performance and adaptive stability.

To achieve these goals, the model employs a multi-objective optimization framework based on Pareto optimality. This framework unifies the processing of four sub-objectives—mean depth, mean torque, depth standard deviation, and torque standard deviation—avoiding the potential bias introduced by subjective weighting of individual objectives. For the specific solution method, the model utilizes an exact full permutation traversal method (exhaustive enumeration). By calculating and comparing all 24 possible parameter combinations, this approach is well-suited for scenarios with a finite solution space and guarantees the identification of the global Pareto-optimal set.

The model first integrates 144 data sets, covering six types of simulated lunar soil and encompassing the full factorial experimental design (3 conicity levels × 4 projected area levels × 2 vane count levels). For each parameter combination, the model extracts depth and torque data across all simulated lunar soil types to calculate the corresponding means and standard deviations. Using non-dominated sorting techniques, the model identifies “non-dominated solutions”—those that are superior in at least one objective without being inferior in any others. The set formed by these solutions constitutes the Pareto frontier.

Finally, the model outputs all solutions located on this frontier (18 solutions identified via the full permutation method). Each solution represents a parameter combination that achieves a distinct optimal balance between “high mean values” and “low volatility” for depth and torque. Based on actual engineering preferences (e.g., prioritizing depth versus prioritizing torque), the final implementation scheme can be selected from this frontier, providing scientific guidance for the optimization design.

3.5. Optimization Results of Cone Tip Dimensions

Based on the response surface results presented in Figure 15, Figure 16 and Figure 17, the effects of cone tip geometric parameters on penetration depth and shear torque exhibit relatively clear patterns across different particle sizes and density states. Overall, penetration depth and shear torque display a certain trade-off relationship: smaller conicities and smaller projected areas are more conducive to enhancing penetration capability. In contrast, larger projected areas are more favorable for obtaining higher torque responses during shearing. Consequently, the key to optimal probe selection does not lie in maximizing a single indicator, but rather in achieving a balance between penetration capability and shear signal strength under limited normal loading conditions.

Figure 15 presents the penetration and shear test results for JLU-M-1 under the loose condition. In the loose JLU-M-1 scenario, the maximum average penetration depth reaches 119.717 mm, corresponding to the parameter combination of a 15° conicity, a 12.83 mm vane diameter, and two vanes; the maximum average torque is 0.26 N·m, corresponding to the combination of a 15° conicity, a 35.7 mm vane diameter, and two vanes. These results indicate that in fine-grained, loose simulant, the penetration resistance is relatively low and the probe can more readily reach greater depths. Therefore, probes with small conicities and small diameters can fully exploit their “ease of penetration” advantage, achieving significantly greater penetration depths. Meanwhile, as the vane diameter increases, the contact area between the probe and the regolith enlarges, and the volume of soil involved in failure and disturbance during shearing grows, resulting in a noticeable increase in torque. This demonstrates that under such conditions, the influence of probe parameters on penetration and shear responses is most sensitive, and a pronounced trade-off exists between the two.

Figure 15 also presents the penetration and shear test results for JLU-M-1 under the dense condition. In the dense JLU-M-1 scenario, the maximum average penetration depth decreases to 38.693 mm, corresponding to the parameter combination of a 45° conicity, a 12.83 mm vane diameter, and two vanes; the maximum average torque is 0.06 N·m, corresponding to the combination of a 30° conicity, a 30 mm vane diameter, and two vanes. Compared with the loose condition, the penetration depth in the dense state is significantly reduced, and the torque is also lower overall, indicating that compaction substantially increases the resistance of the regolith to probe advancement, causing the 16 N normal load limit to be reached more rapidly and resulting in premature termination of penetration. Under these circumstances, small-diameter probes remain beneficial for reducing penetration resistance, while larger-diameter probes, although capable of enhancing the shear response, further compromise the effective penetration depth. This suggests that, in fine-grained, dense regolith, the probe design should prioritize “accessibility” above all else; otherwise, it becomes difficult to ensure the effectiveness of the subsequent shear measurement.

Figure 16 presents the penetration and shear test results for JLU-M-2 under the loose condition. In the loose JLU-M-2 scenario, the maximum average penetration depth is 90.742 mm, corresponding to the parameter combination of a 15° conicity, a 12.83 mm vane diameter, and two vanes; the maximum average torque is 0.123 N·m, corresponding to the combination of a 15° conicity, a 35.7 mm vane diameter, and two vanes. Compared with the loose JLU-M-1 scenario, both the maximum penetration depth and the maximum torque are reduced under this condition, indicating that as particle size increases, the local interlocking and inter-particle meshing effects encountered by the probe during penetration and rotation are intensified, and the overall flowability of the regolith diminishes. Nevertheless, the optimal penetration parameters and optimal torque parameters remain consistent with those observed in the fine-grained loose condition, suggesting that in the loose state, the influence patterns of conicity and projected area on performance exhibit a degree of stability: namely, smaller conicities and smaller diameters are more favorable for penetration, while larger diameters are more conducive to obtaining stronger shear signals.

Figure 16 also presents the penetration and shear test results for JLU-M-2 under the dense condition. In the dense JLU-M-2 scenario, the maximum average penetration depth is 39.153 mm, corresponding to the parameter combination of a 45° conicity, a 12.83 mm vane diameter, and two vanes; the maximum average torque is 0.052 N·m, corresponding to the combination of a 30° conicity, a 35.7 mm vane diameter, and two vanes. Compared with the loose state, the dense state still significantly suppresses penetration depth, and the increase in torque is limited. This indicates that under conditions of medium particle size and relatively high density, the probe is constrained by the load limit and cannot reach sufficient depth, thereby restricting the extent of the subsequent shearing action. This result further demonstrates that in unmanned exploration missions with limited loading capacity, if the probe geometry is oversized, even though it may theoretically contribute to increased shear torque, the overall detection effectiveness may be compromised due to insufficient penetration.

Figure 17 presents the penetration and shear test results for JLU-M-3 under the loose condition. In the loose JLU-M-3 scenario, the maximum average penetration depth is 55.410 mm, corresponding to the parameter combination of a 15° conicity, a 12.83 mm vane diameter, and two vanes; the maximum average torque is 0.08 N·m, corresponding to the combination of a 15° conicity, a 35.7 mm vane diameter, and two vanes. Compared with the two finer particle size conditions, the penetration depth is further reduced under this scenario, indicating that larger particles significantly increase the energy required for soil rearrangement and local crushing ahead of the probe, thereby markedly constraining the penetration capability under limited normal loading. It is noteworthy that the maximum torque also decreases correspondingly, suggesting that coarse-grained regolith does not continuously yield higher torque as a result of “enhanced particle interlocking”; instead, the limited depth of probe entry restricts the full development of the shearing action. Therefore, under coarse-grained loose conditions, insufficient penetration depth has become a critical factor limiting torque output.

Figure 17 also presents the penetration and shear test results for JLU-M-3 under the dense condition. In the dense JLU-M-3 scenario, the maximum average penetration depth is only 32.167 mm, corresponding to the parameter combination of a 30° conicity, a 12.83 mm vane diameter, and two vanes; the maximum average torque is only 0.013 N·m, corresponding to the combination of a 30° conicity, a 30 mm vane diameter, and two vanes. This scenario represents the most unfavorable condition among all six test configurations, demonstrating that coarse-grained and dense simulant imposes the strongest constraints on both probe penetration and shearing. Under these conditions, even increasing the vane diameter fails to significantly enhance torque response, because the probe reaches the 16 N load limit before attaining a sufficient working depth, leaving the subsequent shear test with an inadequate volume of surrounding soil. This result indicates that for extremely coarse-grained, dense regolith, merely enlarging the probe dimensions cannot improve detection performance and may, in fact, further deteriorate penetration capability. Consequently, the probe design must prioritize ensuring penetrability under low-load conditions.

The Pareto-optimal solution set obtained from iterative computation of the full-factorial experimental results is shown in Figure 18, yielding a total of 18 solutions, where each point represents the corresponding depth and torque for different conicities, vane diameters, and vane numbers. Blue triangles denote two-vane probes and red triangles denote four-vane probes.

From the Pareto-optimal solution set and the response surface results, it can be observed that two-vane probes are overall superior to four-vane probes, particularly in terms of penetration depth and comprehensive performance. This indicates that under the constraint of a normal load of only 16 N. However, increasing the number of vanes enlarges the contact area. This simultaneously increases both penetration and rotational resistance, causing the probe to reach the load limit more readily at shallow depths, which is, in fact, detrimental to the acquisition of subsurface information. By contrast, the two-vane probe maintains a certain shearing capability while exhibiting lower entry resistance, and therefore demonstrates better overall adaptability across most parameter combinations. It is thus evident that under the conditions of a lightweight lunar platform, more vanes do not necessarily yield better performance; rather, the number of vanes should be matched to the available loading capacity.

Synthesizing the results across all six test scenarios, a relatively consistent pattern emerges regarding the influence of probe parameters on penetration depth and shear torque. A reduction in conicity is beneficial for enhancing penetration capability, and this effect is particularly pronounced in loose regolith. An increase in vane diameter is favorable for increasing shear torque but comes at the expense of penetration depth. The two-vane configuration is overall superior to the four-vane configuration and is better suited to in situ exploration tasks under the 16 N low normal load constraint. As particle size increases and density rises, the penetration capability and torque response of the probe generally decline, with the dense JLU-M-3 condition being the most unfavorable for testing.

Therefore, if the sole objective is to maximize penetration depth, probes with small conicities and small diameters should be preferentially selected; if the sole objective is to maximize shear torque, probes with larger diameters should be favored. However, for lunar in situ exploration missions, the probe must not only “be capable of obtaining measurements” but also “be capable of penetrating the regolith.” Under this premise, it is not reasonable to pursue the maximization of any single indicator alone; instead, greater emphasis should be placed on achieving a comprehensive balance between penetration capability and shear response.

From a mission design perspective, the results presented here carry important implications for payload sizing on future unmanned lunar rovers. The finding that an oversized probe—while theoretically beneficial for shear torque—may compromise overall detection effectiveness due to insufficient penetration under the limited loading budget, suggests that the conventional approach of maximizing probe dimensions is inappropriate for lightweight platforms. Mission planners should instead adopt a system-level optimization approach that jointly considers the available normal load, the anticipated range of regolith conditions at the target site, and the minimum acceptable signal strength for parameter inversion.

From the Pareto-optimal solution set, the final preferred probe design was selected by assigning equal weighting factors to the two optimization objectives—penetration depth and maximum shear torque. The design that yielded the highest weighted composite score (i.e., the sum of the equally weighted normalized penetration depth and normalized maximum torque) was identified as the optimal configuration. This equal-weight strategy reflects the engineering requirement that both sufficient penetration into the regolith and adequate shear torque magnitude are equally critical for obtaining reliable in situ measurements under the 16 N normal load constraint. On this basis, the Pareto-optimal solution set indicates that the probe with a 15° conicity, a 20.27 mm vane diameter, and two vanes exhibits the best overall performance across all test conditions. This combination corresponds to an average maximum penetration depth of 51.61 mm and an average maximum torque of 0.055 N·m. Compared with small-diameter probes, it is capable of providing a stronger shear response; compared with large-diameter probes, it retains superior penetration capability. It therefore demonstrates more stable adaptability across a variety of simulated regolith conditions. This demonstrates that the combination of a moderate projected area, a smaller conicity, and fewer vanes is better suited to be integrated in situ penetration–shear exploration tasks under the limited loading conditions of future lunar rovers.

Furthermore, to verify the practical applicability of the selected optimal probe under different simulated regolith conditions, the maximum penetration depth and maximum shear torque of the probe with a 15° conicity, a 20.27 mm vane diameter, and two vanes were compiled across all six test scenarios, and the results are presented in Figure 19. As can be seen from Figure 19, this probe exhibits the strongest penetration capability and shear response under the loose JLU-M-1 condition, with a maximum penetration depth of approximately 120 mm and a maximum torque approaching 0.10 N·m, indicating that when the particles are finer and the medium is looser, the probe can both penetrate sufficiently into the medium and generate a pronounced shear resistance response. As the test condition transitions from JLU-M-1 to JLU-M-3 and from the loose state to the dense state, the maximum penetration depth of the probe generally decreases, and the maximum shear torque diminishes correspondingly, demonstrating that increasing particle size and a higher degree of compaction jointly weaken the penetration–shear response capability of the probe.

In the dense state, the increased packing density results in significantly higher penetration resistance and reduced achievable depth under the same normal load constraint. The specific micromechanical mechanisms governing this behavior—such as changes in inter-particle contact distribution and force chain evolution—require further investigation through dedicated experiments (e.g., triaxial compression tests combined with discrete element method simulations) and will be addressed in future work. The maximum torque is likewise lower overall than that in the corresponding loose condition, reflecting the fact that under the 16 N normal load constraint, the probe cannot achieve a sufficient effective penetration depth in the dense medium, thereby diminishing the contact extent and torque output during the shearing phase. In particular, under the dense JLU-M-3 condition, the maximum penetration depth of the probe is only approximately 22 mm and the maximum torque is approximately 0.01 N·m—the lowest values among all test scenarios—indicating that this condition is the most adverse for probe operation. By comparison, under the loose JLU-M-2 and loose JLU-M-3 conditions, the probe can still achieve maximum penetration depths of approximately 60 mm and 50 mm, respectively, while maintaining maximum torques of approximately 0.07 N·m and 0.065 N·m, demonstrating that the selected optimal probe retains satisfactory usability under medium and coarser particle conditions.

The trends reflected in Figure 19 are consistent with the aforementioned Pareto optimization results, confirming that the optimal probe does not achieve the absolute maximum in any single indicator but rather maintains a balance between penetration capability and shear response output across different media conditions, exhibiting favorable overall balance and environmental adaptability. In other words, the reason the combination of a 15° conicity, a 20.27 mm vane diameter, and two vanes emerges as the optimal solution is not that it attains the highest torque or the greatest depth in every scenario, but rather that it consistently achieves the unified goal of “penetrable, shearable, and identifiable” across multiple types of simulated regolith under the low normal load constraint. This renders it more aligned with the practical requirements of probe structural design for lunar in situ exploration missions.

The inferior performance of four-vane probes can be attributed to a coupled mechanical mechanism operating under the constrained normal load. First, the additional vanes increase the total frontal cross-sectional area and soil displacement volume during penetration, elevating the penetration resistance and causing the probe to reach the 16 N load limit at shallower depths. Second, the greater degree of soil disturbance and remolding caused by four vanes partially destroys the natural interlocking fabric of the regolith surrounding the probe, reducing the undisturbed shear strength available for mobilization during torsional shearing. Third, because the four-vane probe achieves less penetration depth, the confining stress at the probe tip location is lower, resulting in diminished shear torque despite the larger total vane surface area. In essence, the marginal gain in shearing surface from additional vanes is more than offset by the compounded losses in penetration depth, confining pressure, and soil structure integrity. This demonstrates that under low normal load constraints, the number of vanes must be matched to the available loading capacity rather than maximized unconditionally.

4. Conclusions

4.1. Key Findings

(1)

An in situ penetration shear apparatus designed for the lunar surface has been successfully developed, with a maximum measurement force of 25 N and a maximum torque of 2.5 N·m. The probe can achieve a penetration depth of 300 mm and rotate 360°. A Python-based PC host software was successfully developed to display and record displacement-force and angular displacement-torque data curves and time parameters, as well as to export field-acquired data. The working principle of the developed penetration shear apparatus is suitable for future unmanned lunar exploration missions.

(2)

The accuracy of the Force-Torque Sensor and motor encoders of the in situ penetration shear apparatus was verified. The R² of the force sensor is 0.99, the R² of the torque sensor is 0.99, the R² of the penetration motor used for depth measurement is 1, and the R² of the shear motor used for angle measurement is 1, demonstrating good working performance.

(3)

Simulated lunar surface conditions were established under different particle sizes. There are three particle size ranges: less than 1 mm, 1 mm–2 mm, and 2 mm–4 mm. Two density states were controlled simultaneously: loose and dense. Probes of different dimensions were adopted, with the following: conicities of 15°, 30°, and 45°; vane diameters of 12.83 mm, 20.27 mm, 30 mm, and 35.7 mm; and vane counts of 2 and 4. Using a full-factorial experimental design, test results for all conditions were obtained. The Pareto-optimal solution set results indicate that the optimal combination is conicity 15°, vane diameter 20.27 mm, and vane count 2. The average maximum penetration depth is 51.61 mm, and the average maximum torque is 0.055 N·m. This optimization result indicates that, under limited normal load (16 N), adopting a probe with a smaller conicity and a medium diameter can effectively balance penetration depth and signal strength, providing theoretical support for the payload design of future missions, such as Chang’e 8. In the future, for simulated lunar soil, modeling will be performed on the mechanical properties of the simulated lunar soil, such as cohesion and internal friction angle, in relation to probe dimensions and experimental data. This will enable in situ prediction of lunar soil mechanical characteristics during the operation of the penetration shear apparatus. The regolith mechanical property data obtained by the probe system could also support in situ resource utilization efforts, particularly by informing excavation and drilling strategies for water extraction from volatile-bearing polar regolith [35].

4.2. Future Work

(1)

It should be noted that the simulants used in this study, while representative of lunar regolith in terms of bulk particle size range, possess several important differences from actual lunar polar regolith that may affect the transferability of the results. First, regarding grain-size distribution, the experiments employed three discrete, narrowly graded particle size fractions to isolate grain-size effects, whereas natural lunar regolith exhibits a continuous, broadly graded distribution with a significant fraction below 20 μm. Second, concerning particle angularity, lunar regolith grains are highly angular and irregular due to the absence of aqueous or aeolian rounding processes, whereas the simulant particles may exhibit comparatively lower angularity depending on their terrestrial formation and processing history. This difference could affect inter-particle interlocking and, consequently, the measured shear resistance. Third, with respect to cohesion, actual lunar regolith exhibits significant apparent cohesion (on the order of 0.1–1 kPa) attributed to electrostatic charging, solar wind implantation, and van der Waals forces in vacuum—effects that cannot be replicated under terrestrial atmospheric conditions. Fourth, lunar polar regions may contain volatile-bearing regolith (e.g., ice-cemented grains) that would substantially alter mechanical behavior compared to the dry simulant used here. These limitations imply that the optimal probe dimensions identified in this study should be validated through future testing with higher-fidelity simulants (e.g., those matching the Apollo-measured gradation curves) and, where possible, under vacuum conditions. For future South Polar missions, the extremely low temperatures in permanently shadowed regions may significantly increase regolith penetration resistance due to ice cementation, posing additional challenges for drilling and probe-based measurement systems [36].

(2)

It should be noted that the present experimental campaign was conducted under terrestrial gravitational conditions (1 g) as a foundational parametric study. On the lunar surface, the reduced gravitational acceleration (approximately 1/6 g) directly affects the available reaction force from the rover platform, the overburden stress profile within the regolith, and consequently the penetration resistance and shear strength mobilization at any given depth. However, the 16 N normal load constraint adopted in this study already accounts for the reduced lunar gravity, as it was derived from the rover’s lunar weight (35 kg × 1.63 m/s² ≈ 57 N) and the associated stability analysis. The relative ranking of probe configurations is expected to remain valid because the dominant factor—the limited normal load budget—is the same constraint that would govern performance on the lunar surface. Nevertheless, future work will include testing under simulated reduced-gravity conditions (e.g., using counterweight systems or parabolic flight experiments) and the development of mechanics-based predictive models that explicitly account for the depth-dependent stress state under lunar gravity, enabling direct extrapolation of ground-based results to lunar operating conditions and further optimization of the apparatus for flight deployment. Additionally, space weathering processes—such as micrometeorite bombardment and solar wind implantation—may modify grain surface characteristics and consequently alter the bulk frictional and cohesive behavior of in situ regolith compared to terrestrial simulants [37,38]. Furthermore, machine-learning approaches have shown promise in predicting mechanical properties of lunar minerals from compositional and structural data [39], offering a potential pathway to complement probe-based in situ measurements with predictive models for rapid geotechnical assessment across diverse lunar terrains.

(3)

Beyond the near-term application to China’s Chang’e lunar exploration program, the in situ geotechnical measurement methodology and optimized probe design presented in this study hold relevance for a range of future lunar surface activities. As international interest in sustained lunar presence intensifies, accurate characterization of regolith mechanical properties will become essential for site selection, foundation engineering, and resource extraction operations. For example, Ahrens et al. identified candidate test mining sites on the lunar surface where knowledge of local soil bearing capacity, shear strength, and compaction state would directly inform excavation strategy and equipment design [40]. Similarly, Leone proposed the Sverdrup–Henson crater near the lunar South Pole as a candidate location for the first permanent settlement [41], where geotechnical surveying of the regolith would be a prerequisite for habitat construction, trafficability assessment, and infrastructure planning. The lightweight, rover-deployable probe system developed in this work offers a practical means of conducting such preliminary geotechnical surveys across multiple candidate sites with minimal payload mass, thereby supporting mission planning for lunar mining, construction, and long-duration habitation activities.