Design and Autonomous Flight Demonstration of a Low-Cost Cardboard-Based UAV

Highlights

What are the main findings?

• The cardboard-based UAV structure sustained the 7.2 kgf design load with strong agreement between experimental results and analytical predictions.
• The custom low-cost avionics (FCC + AHRS) enabled stable autonomous flight, achieving altitude accuracy within ±2 m and path-following errors of approximately ±3 m.

What are the implications of the main findings?

• The proposed platform provides an accessible and low-cost UAV solution for education, rapid prototyping, and research environments.
• It offers a scalable and expendable alternative for disaster response and rapid deployment missions requiring cost-efficiency and operational simplicity.

Abstract

This study presents the development of a low-cost unmanned aerial vehicle (UAV) employing cardboard as the primary structural material and integrating affordable avionics. The wing spar was constructed entirely from cardboard, and its structural performance was experimentally validated through load testing. To address the limitations of low-cost motion sensors, a custom centrifugal acceleration testing device was designed and utilized to evaluate sensor accuracy. Furthermore, an in-house flight control computer (FCC) and avionics suite were developed to achieve low-cost implementation. Flight tests demonstrated that the proposed cardboard-based UAV achieved stable autonomous flight. The findings confirm the feasibility of combining ultra-low-cost structural materials with low-cost avionics, highlighting the potential of this platform for educational and research applications.

1. Introduction

Unmanned Aerial Vehicles (UAVs) have undergone remarkable development over the past decade across military, civil, and academic domains, fundamentally transforming the landscape of modern aerospace engineering. In recent years, increasing attention has been devoted to achieving adequate mission performance through low-cost and lightweight platforms without relying on expensive composite structures or aviation-grade sensors. This paradigm shift has enabled UAVs to play a critical role in diverse fields such as agriculture, disaster response, infrastructure inspection, meteorological observation, and aerospace education, thereby enhancing their accessibility and practical value [1,2,3,4]. Traditionally, UAV development has depended on advanced composite or metallic airframes, precision manufacturing technologies, and high-cost avionics systems [5,6]. While such configurations ensure superior performance and reliability, they inevitably lead to high production and maintenance costs, limiting the accessibility of UAV development and experimentation for small research laboratories and educational institutions. In contrast, recent trends have moved toward the utilization of commercial off-the-shelf (COTS) components and simplified structural materials such as polystyrene, balsa wood or corrugated cardboard, thereby enabling cost-effective fabrication and rapid prototyping of small UAVs [7,8,9].

Such approaches go beyond mere cost reduction; they serve as educationally valuable, hands-on platforms through which students and researchers can directly engage in the full design–fabrication–testing process of flight vehicles, enhancing both technical literacy and experimental capability [10].

The strategic significance of low-cost UAVs has been highlighted in recent military applications. The Corvo Precision Payload Delivery System (PPDS) developed by SYPAQ Systems of Australia, which employs cardboard as its primary structural material, demonstrated tangible operational effectiveness in real-world reconnaissance and logistics missions (Figure 1) [11]. This example clearly illustrates that even disposable or expendable UAVs can yield substantial strategic benefits on the battlefield [12,13,14]. Consequently, in scenarios requiring mass deployment or where attrition is anticipated, low-cost platforms may in fact deliver greater operational efficiency than high-cost systems [15,16,17].

Parallel developments have emerged in the academic and educational sectors. Several institutions, including the University of Hawaii, Kaunas University of Technology, and Inha Technical College, have initiated programs in which low-cost, fixed-wing UAVs constructed from lightweight structural materials are utilized for instruction in flight dynamics, control systems, and composite design (Figure 1) [18,19]. These initiatives provide students of aerospace and mechanical engineering with integrated, experiential learning opportunities, fostering a deeper understanding of aerodynamics, structural mechanics, and avionics integration through practical experimentation.

Nevertheless, the realization of practical low-cost UAVs still faces several technical challenges.

First, the mechanical strength and durability of unconventional structural materials such as cardboard must be validated, particularly in primary load-bearing components such as spars and wings.

Second, low-cost inertial sensors (IMUs) typically suffer from noise and drift, which can compromise flight stability and navigation accuracy; hence, systematic ground-based performance validation and calibration are indispensable [20].

Third, avionics systems designed for low-cost UAVs must achieve a delicate balance between cost efficiency, reliability, and expandability, requiring an integrated design approach encompassing both hardware and software optimization [21].

To address these issues, the present study proposes and experimentally validates the feasibility of a cardboard-based fixed-wing UAV that achieves autonomous flight capability while maintaining minimal cost and structural simplicity.

The scope of this research encompasses three key aspects:

(i)

a structural test of a cardboard box-type spar to evaluate its stiffness and strength under design loads,

(ii)

the development and utilization of a centrifugal acceleration testing apparatus to assess the accuracy of a low-cost inertial sensor, and

(iii)

the design and implementation of a custom, low-cost avionics system, including an in-house flight control computer (FCC) and attitude and heading reference system (AHRS), to realize autonomous flight control.

Unlike previous studies that have typically addressed either low-cost structural design or avionics development in isolation, this work integrates structural validation, sensor evaluation, and autonomous flight demonstration within a unified experimental framework, thereby establishing a coherent methodology for verifying the viability of low-cost UAV platforms.

Furthermore, the findings of this study indicate that such platforms hold substantial potential not only as cost-effective educational tools but also as expendable or semi-disposable aerial systems for logistics, reconnaissance, and disaster-response missions.

Accordingly, this research aims to contribute to the broader discourse on low-cost UAV design as a bridge between educational prototyping and practical field deployment, underscoring its relevance in both academic and operational contexts.

2. Methods

2.1. Design Requirements and Sizing

This cardboard-based UAV targets low cost and reliable autonomous flight. The mission requirements and design assumptions are summarized in Table 1. Based on these inputs, aerodynamic sizing was carried out for a tapered trapezoidal wing with span $\mathrm{b}=1.79 \mathrm{m}$, leading-edge sweep ${\Lambda }_{\le }\approx 30 \mathrm{d}\mathrm{e}\mathrm{g}$, and elevon control. Directional stability is ensured by sizing wing-tip vertical fins via the vertical tail volume coefficient.

Table 1. Design requirements and assumptions (inputs).

Item	Value	Note
Endurance	$50\mathrm{m}\mathrm{i}\mathrm{n}$	Mission
Max speed $V_{max}$	80 km/h (22.2 m/s)	Mission/performance
Payload	1.0 kg	Maximum allowable mission equipment mass (e.g., camera, telemetry module); excludes airframe and propulsion system
Structure mass	2.0 kg	The structure mass of 2.0 kg was derived from a subsystem-level breakdown covering the airframe, propulsion hardware, control mechanisms, avionics, and miscellaneous components (see Table 2).
$\mathrm{M}\mathrm{T}\mathrm{O}\mathrm{W}\mathrm{m}$	3.0 kg (W = 29.43 N)	Assumed
Safety factor $\mathrm{S}\mathrm{F}$	1.2	Structural
Load factor $\mathrm{n}$	2.0	Structural
Stall speed $V_{stall}$	10 m/s	Sizing
$C_{L_{max}}$	1.2	Sizing (conservative)
Air density $\mathsf{\rho }$	1.225 kg/m3	Sea level
Span $\mathrm{b}$	1.79 m	Given
LE sweep ${\Lambda }_{\le }$	30 deg	Wing setup
Taper ratio $\lambda$	0.5	Assumed
Airfoil	MH60 (≈10%)	Reflex, tailless use

Table 1. Design requirements and assumptions (inputs).

Item	Value	Note
Endurance	$50\mathrm{m}\mathrm{i}\mathrm{n}$	Mission
Max speed $V_{max}$	80 km/h (22.2 m/s)	Mission/performance
Payload	1.0 kg	Maximum allowable mission equipment mass (e.g., camera, telemetry module); excludes airframe and propulsion system
Structure mass	2.0 kg	The structure mass of 2.0 kg was derived from a subsystem-level breakdown covering the airframe, propulsion hardware, control mechanisms, avionics, and miscellaneous components (see Table 2).
$\mathrm{M}\mathrm{T}\mathrm{O}\mathrm{W}\mathrm{m}$	3.0 kg (W = 29.43 N)	Assumed
Safety factor $\mathrm{S}\mathrm{F}$	1.2	Structural
Load factor $\mathrm{n}$	2.0	Structural
Stall speed $V_{stall}$	10 m/s	Sizing
$C_{L_{max}}$	1.2	Sizing (conservative)
Air density $\mathsf{\rho }$	1.225 kg/m3	Sea level
Span $\mathrm{b}$	1.79 m	Given
LE sweep ${\Lambda }_{\le }$	30 deg	Wing setup
Taper ratio $\lambda$	0.5	Assumed
Airfoil	MH60 (≈10%)	Reflex, tailless use

Table 2. Subsystem-level structure mass breakdown.

Subsystem	Mass [kg]	Note
Airframe structure	1.20	Wing, fuselage, tail, internal reinforcements
Propulsion components	0.40	Motor, ESC, propeller, power wiring
Control mechanisms	0.15	Flight-control computer (FCC), IMU, receiver, power PCB, wiring harness
Miscellaneous	0.05	Adhesives, tape, small mounts
Total	2.00 kg	Matches structural mass input in Table 1

Table 2. Subsystem-level structure mass breakdown.

Subsystem	Mass [kg]	Note
Airframe structure	1.20	Wing, fuselage, tail, internal reinforcements
Propulsion components	0.40	Motor, ESC, propeller, power wiring
Control mechanisms	0.15	Flight-control computer (FCC), IMU, receiver, power PCB, wiring harness
Miscellaneous	0.05	Adhesives, tape, small mounts
Total	2.00 kg	Matches structural mass input in Table 1

(1)

Wing sizing under the stall constraint

With stall speed $V_{stall}=10 \mathrm{m}/\mathrm{s}$, and $C_{L_{max}}=1.2$

$$L={\displaystyle \frac{1}{2}}\rho V^{2}S{C}_L$$

(1)

where, $m=3.0 \mathrm{k}\mathrm{g}$, $\rho =1.225 \mathrm{k}\mathrm{g}/{\mathrm{m}}^3$ yielding

$$S={\displaystyle \frac{2W}{\rho V_{stall}^2{C}_{L_{max}}}}\approx 0.4{\mathrm{ }\mathrm{m}}^2$$

(2)

Under stall conditions (i.e., $\mathrm{L}=\mathrm{W},\mathrm{V}=V_{stall}, C_L=C_{L_{max}}$).

For a trapezoidal planform with taper ratio $\lambda ={\displaystyle \frac{c_t}{c_r}}$ = 0.5

$$\mathrm{S}={\displaystyle \frac{b}{2}}{C}_r\left(1+\lambda \right),\mathrm{ }\mathrm{ }\mathrm{ }{c}_r={\displaystyle \frac{2S}{b(1+\lambda )}},\mathrm{ }{c}_t=\lambda c_r$$

(3)

so that, $c_r=0.298 \mathrm{m}, c_t=0.149 \mathrm{m}$.

The mean aerodynamic chord (MAC) and its spanwise station are

$$c_{mac}={\displaystyle \frac{2}{3}}{c}_r{\displaystyle \frac{1+\lambda +{\lambda }^2}{1+\lambda }},\mathrm{ }{y}_{mac}={\displaystyle \frac{b}{6}}{\displaystyle \frac{1+2\lambda }{1+\lambda }}$$

(4)

$$c_{mac}\approx 0.232\mathrm{ }\mathrm{m},\mathrm{ }{ y}_{mac}\cong 0.398 \mathrm{m}.$$

With ${\Lambda }_{\le }\approx 30°$, the MAC leading-edge $\mathrm{x}$-location (from the wing-root LE) is

$$x_{\le ,mac}=y_{mac}tan{\Lambda }_{\le }\approx 0.398\times \mathrm{t}\mathrm{a}\mathrm{n}30 \mathrm{deg}\approx 0.230\mathrm{ }\mathrm{m}.$$

At the maximum speed $V_{max}=80 \mathrm{k}\mathrm{m}/\mathrm{h}=22.2 \mathrm{m}/\mathrm{s}$

$$C_L\left(V_{max}\right)={\displaystyle \frac{W}{0.5\rho {V_{max}}^{2}S}}\approx 0.244$$

(5)

which provides ample margin relative to $C_{L_{max}}=1.2$

The wing loading is $\mathrm{W}/\mathrm{S}\approx 73.5 \mathrm{N}/{\mathrm{m}}^2$.

(2)

Airfoil selection (MH60, low-Re performance)

Assuming tailless (elevon) operation, the reflex airfoil MH60 (thickness ≈ 10%) was adopted to secure longitudinal stability and low-speed performance. The geometry of the selected MH60 airfoil is shown in Figure 2. MH60 provides small $\left|C_{m0}\right|$ and benign stall suitable for flying-wing configurations.

With a mean aerodynamic chord of $C_{mac}\approx 0.232 \mathrm{m}$ and design speed range of $10–22.2\mathrm{ }\mathrm{m}/\mathrm{s}$ (stall to maximum operating speed), the corresponding representative Reynolds numbers are approximately

$$\mathrm{R}\mathrm{e}\approx 1.6\times {10}^5\left(\mathrm{a}\mathrm{t}\mathrm{ }\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{l}\mathrm{ }\mathrm{s}\mathrm{p}\mathrm{e}\mathrm{e}\mathrm{d}\right),\mathrm{ }\mathrm{ }\mathrm{R}\mathrm{e}\approx 3.5\times {10}^5\left(\mathrm{a}\mathrm{t}\mathrm{ }\mathrm{m}\mathrm{a}\mathrm{x}\mathrm{i}\mathrm{m}\mathrm{u}\mathrm{m}\mathrm{ }\mathrm{o}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{ }\mathrm{s}\mathrm{p}\mathrm{e}\mathrm{e}\mathrm{d}\right).$$

These values characterize the lower and upper ends of the airfoil’s operating Reynolds-number range. A conservative value of $C_{L_{max}}\approx 1.2$ was used for sizing. The longitudinal stability margin is maintained through elevon trim combined with a CG position set 25–30% of MAC. The final aerodynamic performance is validated in flight tests.

(3)

Wing-tip vertical fins (directional stability)

Directional stability was evaluated using the vertical tail volume coefficient, defined as

$$V_V={\displaystyle \frac{S_V{\mathrm{l}}_V}{Sb}}$$

Thus,

$$S_V={\displaystyle \frac{V_{V}Sb}{{\mathrm{l}}_V}}$$

(6)

where $S_V$ is the total area of the left and right fins, and ${\mathrm{l}}_V$ is the streamwise lever arm from the aircraft center of gravity (CG) to the aerodynamic center of the fin (taken as 25% of the fin MAC). For small fixed-wing UAVs, typical vertical tail volume coefficients lie in the range $V_V=0.02–0.04$. In this study, a mid-range value $V_V=0.03$ was adopted to ensure adequate directional stability while minimizing additional structural weight and drag. An initial geometric estimate of ${\mathrm{l}}_V\approx 0.3 \mathrm{m}$ was obtained from the CAD layout, corresponding to a CG location near 30% of the wing MAC and a fin aerodynamic center located at 25% of its MAC.

Substituting the design wing area and wingspan yields a total required fin area of $S_V\approx 0.0717 {\mathrm{m}}^2$ (combined area of both wing-tip fins). The derived sizing results are summarized in

Table 3. Derived sizing results (outputs; tapered swept wing).

Quantity	Value	Note
Wing area $\mathrm{S}$	0.400 m2	From stall sizing
Root chord $C_r$	0.298 m	$2\mathrm{S}/\left[\mathrm{b}\left(1+\lambda \right)\right]$
Tip chord $C_t$	0.149 m	$\lambda C_r$
MAC $C_{mac}$	0.232 m
MAC span station $y_{mac}$	0.398 m
MAC LE $x_{\le ,mac}$	0.230 m	$y_{mac}tan30°$
$C_L (condition : V_{max}$)	0.244	Adequate margin
Wing loading $\mathrm{W}/\mathrm{S}$	73.5 N/m2	At MTOW
Vertical tail volume $V_V$	0.03 (Target)	Typical 0.02–0.04
Lever arm ${\mathrm{l}}_V$	0.30 m (Initial)	To be confirmed
Total fin area $S_V$	0.0717 m2	L + R; single ≈ 0.036 m2

. The resulting structural configuration and manufactured layout are shown in Figure 3.

This value provides the necessary directional stability margin for the tailless configuration and was later verified in the CAD reference frame and flight-test evaluation.

2.1.1. Cost Considerations

To justify the low-cost objective of this cardboard-based platform, the approximate material and fabrication cost of the prototype was summarized in

Table 4. Airframe material and fabrication cost summary.

Item	Quantity	Unit Price (USD)	Cost (USD)	Note
Cardboard	3 sheet	2.5	7.5	Commercially available
Adhesive (glue)	-	2.0	2.0	Partial usage cost
Laser cutting	-	8.0	8.0	One-time cutting service
Total (airframe)	-	-	17.5	Under USD 20

Note: Costs exclude avionics, propulsion, and battery, as the focus is on demonstrating the low-cost airframe concept.

. The airframe was constructed entirely from commercially available 5-mm cardboard sheets, standard adhesives, and low-cost laser-cutting services. As shown, the total airframe-related cost remained under USD 20, demonstrating that the platform meets the intended affordability target for rapid prototyping and field experimentation.

2.1.2. Estimated Maximum Level-Flight Speed

To verify whether the proposed configuration can plausibly achieve a maximum speed on the order of 90 km/h, a simple thrust–drag balance analysis was performed. The aircraft is powered by an 800-kV brushless out runner motor operating on a 6-cell (6S) Li-Po battery and a 16 × 8-inch fixed-pitch propeller. Bench measurements with this propulsion setup yielded a static thrust of approximately $6 \mathrm{k}\mathrm{g}\mathrm{f} (\approx 59 \mathrm{N})$ at full throttle.

In steady level flight, the required thrust can be approximated by the aerodynamic drag:

$$T_{req}=D\left(V\right)={\displaystyle \frac{1}{2}}\mathrm{ }\rho V^{2}S{C}_D$$

(7)

Using the wing area listed in Table 1 together with a conservative total drag coefficient of $C_D=0.3$, the drag force at 25 m/s (≈90 km/h) is estimated as

$$D(25\mathrm{ }\mathrm{m}/\mathrm{s})\approx 52\mathrm{ }\mathrm{N}\mathrm{ }(\approx 5.3\mathrm{ }\mathrm{k}\mathrm{g}\mathrm{f})$$

(8)

Since the available static thrust (≈59 N) exceeds this required drag, the analysis indicates that a maximum level-flight speed of approximately 23–25 m/s (80–90 km/h) is theoretically achievable for this configuration. This result provides a physically grounded justification for the reported maximum-speed capability, even though detailed airspeed logging was not recorded during the flight tests.

2.2. Structural Design and Testing

2.2.1. Objective

This section addresses a practical question: Can a UAV wing made solely of commercially available 5 mm cardboard sustain the design load derived from the target mission?

We consider an aircraft mass of 2 kg with a 1 kg payload and assume steady coordinated turn at 60 deg bank, yielding a load factor n ≈ 2. With a safety factor SF = 1.2, the design test load corresponds to a central point load of 7.2 kg (converted to force in the calculations). The verification proceeds by (i) obtaining the elastic modulus E from a short-spar bending test using the same cardboard, (ii) using the Euler–Bernoulli deflection formula to predict wing deflection under a central load, and (iii) performing equivalent loading tests on the actual swept wing at feasible fixtures to confirm agreement within a defined tolerance.

Design load is

$$W_d=n·SF·W\cong 70.946 \mathrm{N}(\approx 7.2 \mathrm{k}\mathrm{g}\mathrm{f})$$

(9)

2.2.2. Specimens, Setup, and Measurements

Spar cross-section: rectangular box, outer B × H = 27 × 50 mm, inner b × h = 17 × 40 mm material: cardboard (cardboard, 5T).

Lengths: (A) short spar L = 0.53 m, (B) full-span equivalent L = 1.79 m.

Boundary/loading: simply supported at both ends, central point load; steps 0~7.232 kg.

Measurements: mid-span deflection ${\delta }_{max}$ recorded with a height gauge on a granite surface plate (resolution 0.01 mm); failure mode and occurrence noted.

Material constant. Elastic modulus E obtained by multi-point linear regression of the load–deflection data from the short-spar test.

Reproducibility assurance. All tests followed a standardized loading sequence; each load step was repeated at least twice to confirm linearity and remove outliers. Fixture geometry and boundary conditions were documented via CAD references. Instruments were calibrated before testing; lab temperature was maintained at 22 ± 2 °C. The same cardboard stock was used across specimens to minimize inter-batch variability.

2.2.3. Experimental Validation Under Sweepback Constraint

Because the designed wing incorporates a leading-edge sweep, a direct central point-load test under ideal simply supported conditions could not be performed without inducing torsional deformation or jig interference. Therefore, an equivalent loading approach was adopted, consisting of three sequential steps designed to ensure both analytical consistency and physical plausibility.

(1)

Short-spar test and formula verification.

A straight spar with an effective span of L = 0.53 m was fabricated and tested under a central point load. The load–deflection response was recorded, and the elastic modulus $E$ of the cardboard material was obtained via linear regression.

To verify analytical consistency, the measured mid-span deflection was compared with the classical Euler–Bernoulli beam prediction for a simply supported beam with a center load:

$${\delta }_{theory}={\displaystyle \frac{P{L}^3}{48EI}}$$

(10)

Using the modulus $E$ identified from the load–deflection slope and the measured geometric moment of inertia $I$, the theoretical deflection computed from Equation (10) showed good agreement with the experimental measurement. This confirms that the Euler–Bernoulli model adequately represents the elastic bending behavior of the cardboard spar (Figure 4).

(2)

Partial-wing deflection measurement and rationality assessment.

A representative partial wing section including skins and ribs was tested under the same loading conditions, and the mid-span deflection was recorded. Using the E value obtained from Step 1, the deflection was predicted on the basis of a bare-spar model, applying the same Euler–Bernoulli beam relation referenced in Step 1. As expected, the measured deflection of the skin-integrated structure was smaller than the bare-spar prediction, which is physically reasonable because the skin–rib box configuration increases overall stiffness. (Figure 5).

(3)

Full-span deflection prediction.

Since a direct full-span central loading test was not feasible due to the swept geometry, the elastic modulus E determined in Step 1 and the equivalent second moment of area ${\mathrm{I}}_{\mathrm{e}\mathrm{q}}$ derived from Step 2 were used to predict the full-span deflection. The calculation was performed using the same beam deflection relation previously referenced in Step 1, applied to an equivalent central load of 7.2 kg central load. The resulting maximum deflection was approximately 62 mm, which is within the acceptable elastic limit for the design load case and confirms the structural adequacy of the cardboard wing. (Figure 6).

2.2.4. Results and Discussion

Using the elastic modulus E obtained from the short-spar calibration (Step 1) and the equivalent second moment of area ${\mathrm{I}}_{\mathrm{e}\mathrm{q}}$ derived from the partial-wing test (Step 2), the predicted mid-span deflection of the full-span wing under the equivalent central load of 7.2 kg was approximately δmax ≈ 62 mm. Across all intermediate load steps, the measured load–deflection curves exhibited strong linearity and remained within ±10% of the analytical predictions based on the calibrated E. The maximum bending stress at the design load was approximately 0.76 MPa, which is well below the conservative bending strength limit of the cardboard material (≈6 MPa), indicating a sufficient structural margin.

The comparison between the analytical predictions and the measured deflections under feasible fixture conditions (Figure 5) shows excellent consistency, with the shaded region indicating the ±10% agreement band.

Table 5. Theoretical mid-span deflection (mm) under central point load.

Load (kg)	W (N)	${\mathit{\delta }}_{\mathit{m}\mathit{a}\mathit{x}}\mathit{f}\left(\right)$ Short Spar L = 0.53	${\mathit{\delta }}_{\mathit{m}\mathit{a}\mathit{x}}\mathit{f}\left(\right)$ Full Span L = 1.79
0.000	0.000	0.000	0.000
0.922	9.045	1.143	8.061
1.840	18.050	2.281	16.087
2.724	26.722	3.377	23.815
3.600	35.316	4.462	31.475
4.472	43.870	5.543	39.098
5.376	52.739	6.664	47.003
6.304	61.842	7.814	55.116
7.232	70.946	8.965	63.229

summarizes the theoretical mid-span deflection at each load step; the full-span prediction applies the equivalent moment of inertia ${\mathrm{I}}_{\mathrm{e}\mathrm{q}}$ derived from the partial-wing verification.

3. Avionics System Design and Development

Since the primary goal of this study was the development of a low-cost unmanned aerial vehicle (UAV), cost reduction and weight minimization were key considerations in the design and development of the avionics system. Commercially available flight control computers (FCCs) and attitude and heading reference systems (AHRSs) are generally expensive and therefore unsuitable for small, cost-sensitive UAV platforms.

Accordingly, a custom-designed and fabricated in-house FCC was developed to replace the commercial unit, and a low-cost commercial inertial sensor was adopted as the AHRS to simultaneously achieve cost competitiveness and practical applicability.

3.1. Flight Control Computer (FCC) Development

The flight control computer (FCC) used in this study was designed as a compact, low-cost, and open-architecture system to replace expensive commercial products. The core controller is based on a PJRC Teensy 4.1 microcontroller (MCU) (PJRC, Sherwood, OR, USA), which integrates sensor fusion, control algorithm execution, RC input/output handling, communication, and data logging functions into a single compact board. This design not only enables cost-effective replacement of commercial FCCs but also allows flexible modification and expansion according to future research requirements. The overall hardware configuration of the FCC is shown in Figure 7.

For attitude sensing, a Bosch BNO055 nine-axis inertial measurement unit (IMU) (Bosch Sensortec GmbH, Reutlingen, Germany) was adopted as the AHRS. Because the BNO055 performs quaternion-based orientation estimation internally, it significantly reduces the computational load on the main MCU. A u-blox GNSS receiver was integrated for position and velocity measurements via UART communication. The input/output configuration includes six-channel RC input and six-channel RC output, supporting both PWM and SBUS signal formats on the input side, and providing PWM outputs for servo and ESC control. Additional onboard components include an I²C interface, a buzzer control port, dual 5 V/3.3 V power rails, and reverse-polarity and noise-filtering circuits, ensuring stable operation under field conditions. The FCC firmware was developed in-house using the Arduino-based C/C++ framework(Arduino IDE version 2.3.6). The control software executes an attitude control loop at 200 Hz and a navigation loop at 50 Hz, with real-time scheduling handled by timer interrupts. Communication with the ground control station (GCS) uses a simplified MAVLink-compatible protocol, in which only the necessary portions of the Arduino MAVLink library were selectively implemented in the flight-control computer to minimize code complexity while preserving GCS compatibility. System reliability is further enhanced through watchdog timers and sensor timeout detection mechanisms.

3.2. Application and Limitations of a Low-Cost AHRS

The Attitude and Heading Reference System (AHRS), which is responsible for estimating the attitude of the aircraft, traditionally relied on expensive aviation-grade equipment. However, with recent advancements in low-cost commercial IMU sensors, their performance has significantly improved. These sensors integrate built-in sensor fusion algorithms and are capable of providing relatively stable roll, pitch, and yaw attitude information, as well as angular rate outputs (p, q, r), even under centripetal acceleration conditions.

Nevertheless, under steady turning conditions, such sensors may fail to clearly distinguish between centrifugal acceleration and gravitational acceleration. To verify and address this limitation, the present study developed a ground-based rotational test apparatus that can replicate flight-equivalent conditions.

3.3. Rotational Simulation and Validation of AHRS Performance

3.3.1. Theoretical Background of Steady Turn

In a steady, coordinated turn, the centripetal acceleration ${\mathrm{a}}_{\mathrm{c}}$ generated by the bank angle ϕ is independent of the aircraft’s forward velocity and can be expressed aerodynamically as

$$a_c=\mathsf{g}tan\varphi$$

(11)

From the relation of circular motion, the same acceleration can also be expressed in terms of turn radius r and angular velocity Ω:

$$a_c=r{\Omega }^2$$

(12)

By combining the two equations, the angular velocity required to generate an equivalent centripetal acceleration in steady coordinated turns can be expressed as:

$$\Omega =\sqrt{{\displaystyle \frac{\mathsf{g}tan\varphi }{r}}}$$

(13)

When expressed in revolutions per minute (RPM):

$$\mathrm{R}\mathrm{P}\mathrm{M}={\displaystyle \frac{60}{2\pi }}\sqrt{{\displaystyle \frac{\mathsf{g}tan\varphi }{r}}}$$

(14)

This equation provides the reference rotational speed required for a test apparatus with a given radius to reproduce the same centripetal acceleration experienced by an aircraft at a specific bank angle. The physical interpretation of this relationship is illustrated in Figure 8.

The proposed control system and the specifications of the in-house developed flight control computer (FCC) used in the turning test are summarized in

Table 6. Specifications of the in-house developed flight control computer (FCC).

Category	Specification	Remarks
MCU	PJRC Teensy 4.1	32-bit, 1 MB RAM, 8 MB Flash
Sensor (AHRS)	Bosch BNO055 9-axis IMU	Embedded quaternion computation
GNSS	u-blox GPS	Position, velocity, time data
RC Input	6 channels (PWM/SBUS compatible)	For receiver input
RC Output	6 channels PWM	For servo/ESC control
Interfaces	UART × 4, I2C × 1, USB × 1	For GCS and sensors

.

For a rotational radius of 0.6 m, the required speeds are summarized in

Table 7. Relationship between bank angle vs. equivalent rotational speed for a 0.6 m radius.

Bank Angle (Deg)	10	20	30	45	60
Required Speed (RPM)	16.2	23.3	29.3	38.6	50.8

.

3.3.2. Experimental Setup

The test apparatus was primarily constructed based on a thrust test bench consisting of a propeller, motor, load cell, and torque transducer. While the original purpose of the bench was to measure thrust and torque characteristics of the propulsion system, in this study it was extended by incorporating a rotating shaft structure, enabling both thrust measurement and centrifugal acceleration simulation within a single platform. The overall experimental configuration is illustrated in Figure 9. The Flight Control Computer (FCC) and the Attitude and Heading Reference System (AHRS) were mounted at a radial distance of 600 mm, such that during rotation they experienced centripetal acceleration equivalent to that of an aircraft in steady coordinated turn.

3.3.3. Test Results and Verification

Based on the RPM–bank-angle mapping derived in the previous section, the apparatus was operated at the specified speeds. The attitude outputs from the low-cost AHRS showed excellent agreement with the target bank angles, with deviations within ±2 deg throughout all test cases. This result is illustrated in Figure 10. This result confirms that the ground-based rotational environment successfully replicated the centripetal acceleration characteristics of steady-turn flight. Furthermore, although low-cost motion sensors may exhibit performance degradation under severe vibration (e.g., internal-combustion UAVs), the tested AHRS demonstrated sufficient stability and reliability for use in electric small-scale UAV platforms, where vibration levels are relatively low.

3.4. Summary of Avionics Validation

Through the proposed ground-based rotational testing, the performance of a low-cost AHRS was experimentally validated under flight-equivalent centripetal acceleration. The FCC–AHRS integration successfully maintained accurate attitude estimation within the desired range, confirming its applicability to lightweight, cost-sensitive UAV platforms.

4. Results

4.1. Flight Test Overview

To validate the integrated avionics system of the cardboard drone presented in Section 3, a series of flight tests was conducted under real-world conditions. The primary objective of these tests was to examine whether the fabricated airframe, flight control computer (FCC), and guidance, navigation, and control (GNC) algorithms operated as intended during actual flight. The experiments were carried out in autonomous flight mode, and the main performance metrics evaluated were altitude control accuracy and path-following capability. in all flight tests, the UAV was powered by a simple electric propulsion system consisting of a 800-kV brushless out runner motor, a $16\times 8$ inch fixed-pitch propeller, a 60-A electronic speed controller (ESC), and a 6-cell 24.5 V 10,000-mAh lithium-polymer battery. This configuration provided sufficient thrust margin for takeoff, climb, and cruise while keeping the overall system cost and complexity low, in line with the low-cost design objective of the cardboard-based platform.

4.2. Flight Test Environment

The flight tests were conducted in an open-field environment, where the vehicle was operated in autonomous mode starting from basic takeoff. Navigation relied on standard GPS signals, and flight data were collected in real time through the ground control station (GCS) for subsequent analysis. The flight test sequence is shown in Figure 11.

4.3. Test Results

4.3.1. Altitude Control Performance

Figure 12 shows the altitude response of the aircraft during the test with a target altitude of 150 m. The vehicle reached the commanded altitude in approximately 30 s and then maintained it with small fluctuations around the reference. Although the onboard logging system did not provide sufficient resolution to compute formal statistical measures such as variance over multiple repetitions, the recorded altitude consistently remained within an envelope of about ±2 m from the target across repeated flights.

For reference, the steady-state altitude statistics computed over the 40–160 s interval—mean altitude and standard deviation—are summarized below Figure 12. These results confirm that the PI controller embedded in the FCC operated as intended and that a relatively simple control structure can yield stable and reliable altitude regulation in practice.

4.3.2. Path-Following Performance

Figure 13 presents the trajectory comparison in the X–Y plane. The aircraft completed two laps along the square reference path. Although the onboard GPS logger provided primarily waypoint-based position updates—making precise statistical metrics such as RMS lateral error difficult to compute—the qualitative trajectory comparison shows that the aircraft remained close to the commanded path throughout both laps. The observed lateral deviation, on the order of a few meters, is consistent with the expected performance of standard GPS-based navigation systems.

Moreover, the results demonstrate that the proposed guidance and control architecture—consisting of heading-angle (LOS)–based lateral path guidance, PI controllers for altitude and roll stabilization, and a fixed-speed longitudinal mode without active velocity control—successfully achieved reliable path-following behavior under the given test conditions.

4.3.3. Control Parameters

Table 8. Control algorithm parameters used in the flight-control system.

Control Component	Parameter	Value	Note
Altitude PI Controller	$K_p$	0.80	Tuned to achieve fast climb response without overshoot.
	$K_i$	0.12	Eliminates steady-state altitude error.
Roll PI Controller	$K_p$	1.10	Provides sufficient roll authority for turning.
	$K_i$	0.08	Stabilizes roll angle in gusty conditions.
Heading (Yaw) PI Controller	$K_p$	0.60	Stabilizes course angle under GPS noise.
	$K_i$	0.05	Prevents long-term drift.
LOS Guidance Look-ahead distance	$L$	12 m	Ensures smooth path-following without aggressive turns.

summarizes the parameters used in the PI controllers and the LOS guidance law implemented in the FCC. These parameters were tuned through flight testing to ensure stable altitude, attitude, and path-following performance.

4.3.4. Comparative Evaluation

To contextualize the performance of the proposed cardboard-based UAV, a brief quantitative comparison was made with the SYPAQ PPDS, one of the most widely referenced low-cost cardboard UAV platforms.

Table 9. Brief quantitative comparison with SYPAQ PPDS.

Metric	This Study	SYPAQ PPDS (2023)
Structural material cost	USD 17.5	~USD 700
Structural weight	$2.0 \mathrm{k}\mathrm{g}$	$~5.0 \mathrm{k}\mathrm{g}$
Verified load capability	$7.2 \mathrm{k}\mathrm{g}\mathrm{f}$	$~3 \mathrm{k}\mathrm{g}$ (payload class)
Altitude control accuracy	$\pm 2 \mathrm{m}$	Not reported
Path-following error	$\pm 3 \mathrm{m}$	Not reported
Remarks	Fully autonomous flight	Designed primarily for logistics

summarizes key metrics including structural material cost, structural weight, load capability, and autonomous flight performance. The proposed UAV demonstrates significantly lower material cost while achieving verified altitude and path-following accuracy—performance metrics that have not been quantitatively reported for the PPDS.

The comparison highlights that the proposed UAV achieves a substantially lower material cost while maintaining structural reliability and demonstrating autonomous flight performance superior to other low-cost cardboard platforms. Notably, although the SYPAQ PPDS is operationally relevant [11,12], no quantitative autonomous control results have been reported in the literature, whereas the present study provides experimentally validated altitude and lateral tracking accuracy. This comparison supports the contribution of the proposed system as a practical, ultra-low-cost autonomous UAV platform.

Furthermore, the flight stability characteristics observed in the present work confirm that the platform is capable of supporting classical PID controllers, as widely applied in low-cost UAVs [1,22,23]. Given the onboard MCU resources, sensor update rates, and demonstrated control bandwidth, the platform is also compatible with more advanced algorithms such as gradient-free or cooperative source-seeking strategies reported in the literature [24,25]. These capabilities indicate that the proposed cardboard UAV is not only cost-effective but also suitable as a research testbed for both classical and advanced control methodologies [26,27].

5. Conclusions

This study presented a step-by-step validation of a low-cost unmanned aerial vehicle (UAV) system constructed entirely from commercially available components and cardboard-based materials. The overall research was structured into three sequential phases—structural reliability, sensor reliability, and flight performance—in order to demonstrate the engineering feasibility and practical applicability of a lightweight, cost-effective aerial platform.

First, through the cardboard spar structural tests, it was experimentally verified that cardboard structures can withstand the target load of 7.2 kg while maintaining low weight and ease of fabrication when an appropriate cross-section design and manufacturing process are applied. These findings confirm that even inexpensive materials can provide the fundamental structural reliability required for educational and experimental airframes.

Second, to evaluate the limitations of a low-cost AHRS (BNO055-based), a ground-based rotational test apparatus was developed to reproduce centripetal acceleration environments equivalent to steady-turn flight conditions. This setup enabled quantitative verification of the sensor’s ability to distinguish gravitational and centrifugal accelerations.

The AHRS demonstrated stable attitude estimation with deviations within ±2 deg, thereby experimentally proving the practical reliability of low-cost motion sensors—an aspect rarely addressed in previous low-cost UAV research.

Third, based on these structural and sensor validations, autonomous flight tests were conducted. The results showed altitude-holding accuracy within ±2 m and path-following errors of approximately ±3 m, which are consistent with the expected performance of standard GPS-based navigation. These outcomes confirm that the proposed system achieves stable flight control within the inherent limitations of low-cost navigation hardware.

In conclusion, this study demonstrated that a fully functional UAV can be realized using only cardboard materials and low-cost commercial sensors, while ensuring reproducible structural strength and reliable sensor performance.

Beyond serving as an educational prototype, the developed platform can be extended as a versatile, low-cost experimental testbed for validating structural feasibility and flight-control algorithms.

Future research will focus on integrating vision-based autonomous navigation to enable robust attitude and position estimation in GNSS-denied environments, thereby advancing toward self-contained, low-cost autonomous flight systems.