Robotics

22 pages, 54738 KB

Open AccessArticle

Approximate C-Uniform Sampling: An Information-Theoretic and Bayesian Inference Perspective

by Timur Akhtyamov, German Devchich, Egor Kuznetsov, Denis Fatykhov, Chengpu Yu and Gonzalo Ferrer

Robotics 2026, 15(7), 139; https://doi.org/10.3390/robotics15070139 - 22 Jul 2026

Viewed by 510

Abstract

Sampling control trajectories from standard distributions—a foundation for Model Predictive Control (MPC) and Model Predictive Path Integral (MPPI) methods—although probabilistically complete, in practice leads to poor exploration of the configuration space, resulting in catastrophic outcomes in the robot’s operation. Recent works have proposed [...] Read more.

Sampling control trajectories from standard distributions—a foundation for Model Predictive Control (MPC) and Model Predictive Path Integral (MPPI) methods—although probabilistically complete, in practice leads to poor exploration of the configuration space, resulting in catastrophic outcomes in the robot’s operation. Recent works have proposed a family of approaches for sampling in control space that lead to uniform coverage of the configuration space, introducing the notion of C-Uniformity. However, those methods suffer from the need for state and action space discretization and level-sets, pre-computation. In this work, we introduce a proof-of-concept training strategy for deep generative models to achieve diverse and near C-Uniform sampling capabilities. Two variational inference approaches—information maximization algorithm and Stein variational gradient descent—are used as a foundation for training normalizing flow and flow matching models to sample control sequences that lead to a wider coverage of the configuration space, compared to the standard distributions, without state/action space discretization. Qualitative and quantitative evaluations support the advantage of the methods in terms of state space coverage and success rate in downstream social navigation tasks. Full article

(This article belongs to the Section AI in Robotics)

► Show Figures

Figure 1

25 pages, 21883 KB

Open AccessArticle

Safety-Critical Coverage Control-Based Trajectory Generation for Formation Control of UAV Systems

by Zahra Kashi, Mohammadhasan Faghihi, Nargess Sadeghzadeh-Nokhodberiz and Allahyar Montazeri

Robotics 2026, 15(7), 138; https://doi.org/10.3390/robotics15070138 - 22 Jul 2026

Viewed by 531

Abstract

This paper presents a safe trajectory generation framework for formation production in Multi-Agent Systems (MASs). The proposed approach integrates coverage control with Control Barrier Functions (CBFs) to generate safe trajectories that guide agents from arbitrary initial positions to a desired formation. The environment [...] Read more.

This paper presents a safe trajectory generation framework for formation production in Multi-Agent Systems (MASs). The proposed approach integrates coverage control with Control Barrier Functions (CBFs) to generate safe trajectories that guide agents from arbitrary initial positions to a desired formation. The environment is partitioned into informative and non-informative regions, where the informative regions characterize the desired geometric structure, which effectively defines the final formation. A coverage-control-based algorithm first generates nominal trajectories that drive agents toward these informative regions. To ensure safety, CBFs are incorporated as a supervisory layer that enforces obstacle avoidance and inter-agent collision avoidance while minimally modifying the nominal control inputs. Consequently, the resulting trajectories remain safe while preserving convergence toward the desired formation. Full article

(This article belongs to the Special Issue Advanced Control and Optimization for Robotic Systems)

► Show Figures

Figure 1

26 pages, 9377 KB

Open AccessArticle

Motion-Aware Autonomous Exploration Framework for AUVs in Complex Underwater Structures

by Shihui Shen, Kanghui Jiang, Mingyang Dai and Hongzhi Wang

Robotics 2026, 15(7), 137; https://doi.org/10.3390/robotics15070137 - 21 Jul 2026

Viewed by 416

Abstract

The autonomous exploration capability of Autonomous Underwater Vehicles (AUVs) in complex underwater structures is fundamentally constrained by the coupling between sensing and motion execution. Unlike ground and aerial robots, limited underwater communication makes it difficult for AUVs to rely on external computing resources [...] Read more.

The autonomous exploration capability of Autonomous Underwater Vehicles (AUVs) in complex underwater structures is fundamentally constrained by the coupling between sensing and motion execution. Unlike ground and aerial robots, limited underwater communication makes it difficult for AUVs to rely on external computing resources for perception and motion computation, imposing higher requirements on algorithmic efficiency. Meanwhile, motion constraints require arc-shaped course adjustments rather than in-place turns, thereby increasing navigation costs. This paper focuses on fixed-depth two-dimensional exploration represented by an occupancy grid and assumes deterministic sonar observations in order to isolate the effect of sensing–motion coupling. This sensing–motion coupling makes conventional frontier-based exploration strategies inefficient for underwater environments. To address this issue, this paper proposes a motion-aware autonomous exploration framework for AUVs that jointly considers information acquisition and motion cost during exploration decision-making and path execution. The proposed framework constructs adaptive local viewpoint sampling regions directly from the real-time sonar detection coverage and introduces a weighted sampling strategy to improve viewpoint generation efficiency in structured underwater environments. A motion-aware viewpoint utility function is further designed by integrating frontier information gain, path cost, and heading deviation, enabling the exploration strategy to favor viewpoints with lower execution cost. To balance exploration efficiency and coverage completeness, a local-priority and global-backtracking target assignment mechanism is developed. In addition, motion-constrained path execution is achieved through Douglas–Peucker keypoint extraction, cubic Bezier path smoothing, and dynamic window approach (DWA)-based local trajectory tracking. Experiments on a high-fidelity Unreal Engine–ROS simulation platform show consistent reductions in exploration path length and mapping time relative to a classical frontier-based baseline, with average improvements of about 11–15% in the tested scenarios. In corridor scenarios, the adaptive viewpoint generation strategy improves local viewpoint generation efficiency by 58.5% compared with a sliding-window RRT baseline. The results demonstrate that the proposed framework can effectively improve exploration efficiency and motion consistency for AUV autonomous exploration in complex underwater structures. Full article

(This article belongs to the Special Issue SLAM and Adaptive Navigation for Robotics)

► Show Figures

Figure 1

17 pages, 20231 KB

Open AccessArticle

SAR-SLAM: Semantic-Aware Recognition for Dynamic SLAM in Robotic Applications

by Basheer Al-Tawil, Magnus Jung, Thorsten Hempel and Ayoub Al-Hamadi

Robotics 2026, 15(7), 136; https://doi.org/10.3390/robotics15070136 - 20 Jul 2026

Viewed by 377

Abstract

Simultaneous Localization and Mapping (SLAM) is essential for autonomous systems navigating in human-centric environments, yet conventional systems fail when people and objects move through the scene. This paper introduces SAR-SLAM (Semantic-Aware Recognition SLAM), an RGB-D SLAM framework that robustly handles dynamic scenes containing [...] Read more.

Simultaneous Localization and Mapping (SLAM) is essential for autonomous systems navigating in human-centric environments, yet conventional systems fail when people and objects move through the scene. This paper introduces SAR-SLAM (Semantic-Aware Recognition SLAM), an RGB-D SLAM framework that robustly handles dynamic scenes containing moving people and objects using dual semantic geometric processing. First, we employ YOLOv8-based semantic segmentation to identify dynamic objects and generate initial detection masks. Second, we apply RANSAC-based Homography analysis to perform geometric motion verification, distinguishing truly moving objects from stationary ones by analyzing feature correspondence patterns. Third, an adaptive fusion mechanism combines both semantic and geometric evidence while incorporating temporal consistency and coverage constraints to maintain system stability. The system is implemented as a modular ROS2 package, enabling smooth integration with robotic systems and compatibility with existing navigation frameworks. SAR-SLAM reduces Absolute Trajectory Error by up to 96% over ORB-SLAM3 on the dynamic sequences of the TUM RGB-D benchmark, and remains competitive with state-of-the-art dynamic SLAM methods across a range of dynamic scenarios. Full article

(This article belongs to the Special Issue Localization and 3D Mapping of Intelligent Robotics)

► Show Figures

Figure 1

25 pages, 21270 KB

Open AccessArticle

Vision-Language Model-Guided Transparent Object Perception and Task-Oriented Grasping for Robotic Manipulation

by Kejian Ni, Xiepeng Yang, Tao Chen and Minglu Zhu

Robotics 2026, 15(7), 135; https://doi.org/10.3390/robotics15070135 - 16 Jul 2026

Viewed by 506

Abstract

Transparent objects such as glass containers, test tubes, and plastic bottles are common in robotic manipulation scenarios, but their refractive and reflective surfaces produce incomplete RGB-D geometry and make task-specific grasp selection unreliable. This paper presents an integrated vision-language system for transparent object [...] Read more.

Transparent objects such as glass containers, test tubes, and plastic bottles are common in robotic manipulation scenarios, but their refractive and reflective surfaces produce incomplete RGB-D geometry and make task-specific grasp selection unreliable. This paper presents an integrated vision-language system for transparent object perception and task-oriented grasping. First, we construct VLM-DRE, a transparent object image instruction dataset with 12,700 images and 38,100 image-instruction-bounding-box triplets. LoRA fine-tuning of Molmo-7B improves target click accuracy from 86.4% to 91.5% and IoU@0.75 from 57.5% to 69.1%. Second, MSR-Net performs monocular depth completion and mask prediction using multi-scale adaptive feature fusion and progressive feature refinement, achieving RMSE 0.066, mAP 98.61%, and IoU 94.12% on Syn-TODD, and RMSE 0.118, mAP 99.02%, and IoU 87.95% on ClearPose. Third, LMF-Net combines RGB-D cross-modal fusion with learnable multi-factor matching to rank AnyGrasp 6-DoF candidates, reaching 77.8% Top-1 and 90.5% Top-3 accuracy on TaskGrasp-Image and improving PRISM-Real success from 61.1% to 68.5%. On a RealSense D435i–Unitree Z1 Pro platform, the complete system obtains 85.4% success with manual clicks and 71.3% with VLM-predicted clicks, supporting perception-to-grasping integration while highlighting target localisation and runtime as deployment bottlenecks. Full article

(This article belongs to the Section AI in Robotics)

► Show Figures

Figure 1

32 pages, 16609 KB

Open AccessArticle

Immersive AR–ROS 2 Teleoperation Architecture for a Physical Quadruped Robot

by Erick Criollo, William Oñate, Gustavo Caiza, Víctor H. Andaluz and José Varela-Aldás

Robotics 2026, 15(7), 134; https://doi.org/10.3390/robotics15070134 - 15 Jul 2026

Cited by 1 | Viewed by 565

Abstract

Immersive teleoperation of quadruped robots requires the operator to interpret a remote environment, make decisions, and maintain control over a dynamic platform through mediated visual feedback and networked command transmission. This study presents and validates a reproducible augmented reality (AR)–ROS 2 architecture designed [...] Read more.

Immersive teleoperation of quadruped robots requires the operator to interpret a remote environment, make decisions, and maintain control over a dynamic platform through mediated visual feedback and networked command transmission. This study presents and validates a reproducible augmented reality (AR)–ROS 2 architecture designed to analyze the relationship between system-level technical conditions and operator experience during immersive teleoperation of a physical quadruped robot. The system integrates Meta Quest 3, Unity, UDP communication, ROS 2 middleware, and Xiaomi CyberDog within a modular workflow that jointly supports immersive control, live RGB/depth perception, physical robot execution, and user-centered HRI evaluation. The architecture decouples command transmission and visual feedback into two UDP-based channels: the control channel maps virtual joysticks and discrete HUD buttons to ROS 2 locomotion and action commands, while the perception channel transmits compressed RGB and depth frames for display in the AR interface. Under nominal conditions, the control channel achieved a mean end-to-end latency of 9.37 ms, a P95 of 18.45 ms, an effective update frequency close to 18 Hz, and 100% packet reception. Compared with ROS-TCP-Endpoint, the proposed UDP bridge showed similar latency but higher flow integrity, avoiding duplicate, parsing, and dropped-command events. The visual channel achieved mean end-to-end latencies of 21.09 ms for RGB and 26.99 ms for depth, with frame reception rates of 91.78% and 91.10%, respectively, under a shared 18.67 Mbps mobile network. A complementary network degradation analysis showed that the control channel remained operational under bandwidth reduction, packet loss, added latency, combined degradation, and physical separation of the robot. The user evaluation with 25 participants yielded a corrected Raw NASA-TLX score of

7.13 \pm 4.10

, a positive perceived performance score of

95.20 \pm 5.10

, and a System Usability Scale score of

90.30 \pm 6.63

, indicating low perceived workload and high usability. These results show that low end-to-end latency, high flow integrity, and stable visual feedback were accompanied by low perceived workload and high usability, indicating that the technical temporal response, stability, and robustness of the communication channels are consistent with an effective and low-demand operator experience during AR-based teleoperation of a physical quadruped robot. Full article

(This article belongs to the Special Issue Legged Robots into the Real World, 3rd Edition)

► Show Figures

Figure 1

24 pages, 1967 KB

Open AccessArticle

Safety-Governed Development of a Pediatric Robotic Elbow Orthosis for Arthrogryposis Multiplex Congenita: A Multi-Standard Case Study in an Academic Resource-Constrained Setting

by Alberto Isaac Pérez-Sanpablo, Alicia Meneses-Peñaloza, Citlalli Jessica Trujillo-Romero, Santos M. Orozco-Soto, Lorena Parra-Rodríguez, Montserrat Godínez-García, Aldo R. Mejía-Rodríguez, Marcela D. Rodríguez, José Ambrosio-Bastián and Zizilia Zamudio-Beltrán

Robotics 2026, 15(7), 133; https://doi.org/10.3390/robotics15070133 - 13 Jul 2026

Viewed by 707

Abstract

Robotic systems for pediatric rehabilitation must provide precise mechanical assistance while ensuring clinically appropriate risk control for vulnerable populations. In low- and middle-income countries (LMICs), academic medical robotics projects frequently fail to progress beyond intermediate Technology Readiness Levels (TRLs 3–5) due to limited [...] Read more.

Robotic systems for pediatric rehabilitation must provide precise mechanical assistance while ensuring clinically appropriate risk control for vulnerable populations. In low- and middle-income countries (LMICs), academic medical robotics projects frequently fail to progress beyond intermediate Technology Readiness Levels (TRLs 3–5) due to limited translational planning. This study proposes and evaluates an integrated governance framework for academic pediatric rehabilitation robotics in LMIC settings, applied through the development of the AMCOR robotic orthosis for pediatric arthrogryposis multiplex congenita (AMC). The framework combines multiple national and international medical devices development standards and a dual regulatory pathway separating academic development from future translational stages. The framework is structured around four principles—auditability, TRL-proportional documentation, binding decision criteria, and regulatory separation—and is operationalized through a six-gate process. Across the first two gates (G0–G1), 38 traced requirements and 12 failure modes were documented. Five internal audits confirmed operational implementation of the quality management structure. The framework application also shaped core engineering decisions. The low amplitude and poor signal-to-noise ratio of sEMG signals observed in pediatric AMC patients rendered the original single-layer control strategy inadequate, prompting a framework-governed redesign toward a three-layer adaptive architecture based on signal quality thresholds and fallback safety logic. These findings demonstrate that a prospective, multi-standard governance model can improve early-stage academic medical robotics in resource-constrained settings. Generalizability beyond the single-center, two-gate application reported here requires further validation; however, the framework provides a replicable foundation for adoption in comparable LMIC contexts. Full article

(This article belongs to the Section Medical Robotics and Service Robotics)

► Show Figures

Figure 1

21 pages, 10497 KB

Open AccessArticle

Ray-Casting-Based Trajectory Generation for Industrial Robots in Manufacturing Operations

by Eduardo Fuentes-Fierro, Erardo Leal-Muñoz and Eduardo Diez

Robotics 2026, 15(7), 132; https://doi.org/10.3390/robotics15070132 - 10 Jul 2026

Viewed by 477

Abstract

This paper proposes a set of trajectory generation strategies for industrial robots that use ray-casting over the workpiece CAD model for various manufacturing operations. By employing ray-casting on a triangular-mesh representation of the production part, points can be generated across the entire surface [...] Read more.

This paper proposes a set of trajectory generation strategies for industrial robots that use ray-casting over the workpiece CAD model for various manufacturing operations. By employing ray-casting on a triangular-mesh representation of the production part, points can be generated across the entire surface without extracting geometric features such as curves, edges, or planes. This approach enables the development of diverse point-generation methods with distinct characteristics, adaptable to the specific requirements of each part and manufacturing process. The developed algorithms achieve results comparable to existing robot programming methods, and, when integrated into the specialized offline programming environment, they enable flexible trajectory generation for operations such as sanding, milling, adhesive deposition, and painting. Finally, these trajectories are automatically exported in a syntax that ensures rapid integration of the point sequence into a base program compatible with an articulated robot controller. The results show that the proposed methods can effectively generate trajectories for sanding and milling using two different robots. Full article

(This article belongs to the Topic Smart Production in Terms of Industry 4.0 and 5.0)

► Show Figures

Figure 1

21 pages, 10156 KB

Open AccessArticle

ROS2-Based Low-Cost Mobile Robot for Educational Assistance with Reactive Navigation and Semantic-Cached Language Processing

by Sebastián Alexis Aucapiña, Nataly Cecilia Benalcázar, José Varela-Aldás and Ramiro Isa-Jara

Robotics 2026, 15(7), 131; https://doi.org/10.3390/robotics15070131 - 8 Jul 2026

Viewed by 653

Abstract

Educational environments, particularly those with limited resources, require affordable mobile robots capable of combining human–robot interaction, autonomous assistance, and academic support without continuous dependence on cloud services. This work presents a low-cost ROS2-based mobile robot implemented on a Raspberry Pi 4B to provide [...] Read more.

Educational environments, particularly those with limited resources, require affordable mobile robots capable of combining human–robot interaction, autonomous assistance, and academic support without continuous dependence on cloud services. This work presents a low-cost ROS2-based mobile robot implemented on a Raspberry Pi 4B to provide educational assistance in Spanish within controlled classroom environments. The system integrates voice interaction, text-to-speech synthesis, YOLOv8n-based object perception, a specialized door detection model, ultrasonic and inertial sensing, differential-drive control, and a hybrid natural language processing architecture based on semantic caching, local inference, and optional cloud connectivity. Two task-dependent operating modes, education and navigation, selectively activate ROS2 nodes to reduce computational load and energy consumption. Experimental tests conducted in a university classroom evaluated speech recognition, vision models, natural language processing alternatives, sensor behavior, and battery life. The speech recognition module achieved 98% accuracy under both quiet and noisy conditions. YOLOv8n achieved an F1-score of 0.975 for common classroom objects, while the specialized door detector achieved 100% recall with 58.7% precision. The semantic cache correctly resolved recurrent academic queries in the exact-match evaluation, with an average latency of 3.8 s, reducing the need for external language models in known-question scenarios. The robot operated for 96 min in education mode and 75.6 min in navigation mode. These results demonstrate that Spanish voice interaction, reactive navigation, academic question answering, and resource-aware operation can be integrated into a single low-cost edge robotic platform for educational environments. Full article

(This article belongs to the Section Educational Robotics)

► Show Figures

Figure 1

17 pages, 8960 KB

Open AccessArticle

Experimental Validation of ASSIST-FEEv3 Elbow Assisting Device with Physiotherapy Considerations

by Cuauhtémoc Morales-Cruz, Fortunato Frisina, Francesco Scerbo, Rocco Mazzotta and Marco Ceccarelli

Robotics 2026, 15(7), 130; https://doi.org/10.3390/robotics15070130 - 3 Jul 2026

Viewed by 534

Abstract

Upper-limb rehabilitation and elderly exercise programs require lightweight, reliable, and physiotherapy-oriented assistive technologies capable of supporting controlled joint motion while enabling objective performance assessment. This paper presents experimental validation of ASSIST-FEEv3, a cable-driven elbow assisting device that is designed for flexion–extension exercises with [...] Read more.

Upper-limb rehabilitation and elderly exercise programs require lightweight, reliable, and physiotherapy-oriented assistive technologies capable of supporting controlled joint motion while enabling objective performance assessment. This paper presents experimental validation of ASSIST-FEEv3, a cable-driven elbow assisting device that is designed for flexion–extension exercises with emphasis on usability, portability, and physiotherapy integration. The device employs a dual-cable antagonistic mechanism that is actuated by servomotors housed in a compact module, allowing guided motion in the arm sagittal plane with minimal wearable load mass. A testing campaign was conducted with 25 healthy volunteers under the supervision of physiotherapy experts following a properly designed protocol for three sessions of ten repetitions each. Joint kinematics was acquired through integrated sensing, and performance metrics including maximum flexion, maximum extension, and range of motion (ROM) were analyzed to assess repeatability, motion smoothness, and user-specific variability. The results demonstrate consistent motion assistance across repeated cycles, variability between sessions, and comparable ROM distributions between sexes. Observed deviations were considered due to individual temporary conditions rather than device-related limitations. The device operated with low energy consumption as required in home-based applications. Test findings validate both the mechanical reliability and the physiotherapy-oriented operational framework of the ASSIST-FEEv3 device. Full article

(This article belongs to the Special Issue Selected Papers from MESROB2025: 9th IFToMM International Workshop on New Trends in Medical and Service Robotics)

► Show Figures

Figure 1

27 pages, 8168 KB

Open AccessArticle

Flexipede: A Bio-Inspired, Modular Myriapod Robot for Rough-Terrain Traversal

by Samudra Jit Saha, Md. Abid Chowdhury, Sayma Islam, Shamim Ahmed Deowan, Shifat E. Arman and Abhishek K. Ghosh

Robotics 2026, 15(7), 129; https://doi.org/10.3390/robotics15070129 - 1 Jul 2026

Viewed by 867

Abstract

Rough-terrain exploration is critical for applications ranging from post-disaster search-and-rescue to planetary exploration. While conventional wheeled or bipedal robots often struggle in these environments, biological organisms like myriapods demonstrate superior adaptability. Inspired by this, we present Flexipede—a compact, modular robotic system that employs [...] Read more.

Rough-terrain exploration is critical for applications ranging from post-disaster search-and-rescue to planetary exploration. While conventional wheeled or bipedal robots often struggle in these environments, biological organisms like myriapods demonstrate superior adaptability. Inspired by this, we present Flexipede—a compact, modular robotic system that employs a hybrid actuation architecture, wherein each module integrates a single actuator for propulsive gait generation and a secondary actuator to enable distributed yaw control. The platform is fully 3D-printable and cost-effective, with a fabrication cost of approximately $58 for the primary unit and $10 per additional module. Analytical kinematic modeling was employed to optimize linkage trajectories, with experimental results validating the system across six modular configurations and three distinct environments, including flat, rough, and inclined terrains. The platform achieved locomotion speeds up to 9 cm/s and navigated obstacles up to 32 mm high, while linkage path deviations remained functionally negligible (mean deviation of

3.24 %

) compared to analytical prediction. Turning performance converged with theoretical predictions as modular scaling increased, reaching a minimum deviation of

4.51 %

for the five-module configuration. Across all terrains, the system maintained a competitive average Cost of Transportation (CoT) of

18.01

, with stair climbing requiring a relatively higher CoT due to the elevated torque demands associated with vertical displacement. These results establish Flexipede as a high-performance benchmark for modular myriapod systems with significant potential for adaptive morphological research. Full article

(This article belongs to the Section Intelligent Robots and Mechatronics)

► Show Figures

Figure 1

24 pages, 8403 KB

Open AccessArticle

Depth-Assisted Sparse Visual Odometry for UAV-Relevant Synthetic RGB-D Evaluation: A Controlled Geometric-Backend Ablation

by Andrii Polukhin, Sergii Stirenko, Mairo Leier, Gert Jervan, Oleksandr Rokovyi, Oleg Alienin, Nazrul Nazeer and Yuri Gordienko

Robotics 2026, 15(7), 128; https://doi.org/10.3390/robotics15070128 - 30 Jun 2026

Viewed by 563

Abstract

Sparse visual odometry (VO) is a core component of lightweight unmanned aerial vehicle (UAV) visual navigation, yet the isolated effect of adding aligned metric depth to a minimal frame-to-frame pipeline is easily obscured in full SLAM systems. This paper presents a UAV-relevant controlled [...] Read more.

Sparse visual odometry (VO) is a core component of lightweight unmanned aerial vehicle (UAV) visual navigation, yet the isolated effect of adding aligned metric depth to a minimal frame-to-frame pipeline is easily obscured in full SLAM systems. This paper presents a UAV-relevant controlled synthetic ablation of RGB-only and RGB-D geometric backends under fixed sparse frontends. ORB matching and KLT tracking are evaluated on a 32-sequence TartanAir validation split of flight-like synthetic RGB-D scenes by routing identical 2D correspondences either to Essential Matrix estimation or, with aligned depth, to PnP with RANSAC. The study reports ATE, Sim(3)-aligned ATE, translational and rotational RPE, robustness under temporal subsampling and RGB degradations, and isolated solver latency on a workstation and Raspberry Pi 4. At stride 1, RGB-D PnP reduces ATE by 61.8% for KLT and 29.1% for ORB, with translational RPE reductions of 61.6% and 41.9%. Rotational RPE reductions are stronger and persist across all tested strides, reaching 85.9% for KLT and 77.3% for ORB at stride 1. Sim(3) analysis shows that only 7–16% of PnP ATE is metric-scale drift. At coarser strides, however, KLT-PnP no longer improves ATE, showing that depth assistance depends on stable frontend tracking and valid depth-supported correspondences. The contribution is a reproducible diagnostic benchmark and failure-mode analysis for UAV-relevant depth-assisted sparse VO under oracle aligned depth, providing component-level evidence rather than full onboard deployment validation. Full article

(This article belongs to the Special Issue UAV Systems and Swarm Robotics: 2nd Edition)

► Show Figures

Figure 1

24 pages, 5385 KB

Open AccessArticle

A Late-Fusion Multimodal Approach for Safety-Aware Workspace Modeling in Collaborative Robotic Systems

by Kevin David Ortega-Quiñones, Elias Escobar-Pereira, Michael Felipe Cifuentes-Molano, Germán Andrés Holguín-Londoño and Mauricio Holguín-Londoño

Robotics 2026, 15(7), 127; https://doi.org/10.3390/robotics15070127 - 30 Jun 2026

Viewed by 411

Abstract

Ensuring safe coexistence between human operators and industrial robot manipulators is a critical challenge in collaborative manufacturing environments. Existing approaches rely either on dedicated safety-rated hardware, which is expensive and difficult to retrofit, or on purely vision-based classifiers that discard the precise kinematic [...] Read more.

Ensuring safe coexistence between human operators and industrial robot manipulators is a critical challenge in collaborative manufacturing environments. Existing approaches rely either on dedicated safety-rated hardware, which is expensive and difficult to retrofit, or on purely vision-based classifiers that discard the precise kinematic state available from the robot controller, leading to unresolved visual ambiguities when different joint configurations produce similar appearances from fixed camera viewpoints. Kinematics-only approaches, while precise, lack the spatial context needed to disambiguate configurations near workspace boundaries. We propose RGBJointsNet, a late-fusion multimodal deep learning classifier that combines RGB visual features extracted by a frozen EfficientNet-B2 convolutional backbone with a compact kinematic stream processing the 12-dimensional joint angle vector of a dual-UR5 robotic cell. The model maps each observation to one of five mutually exclusive workspace zones: rest (

C_{0}

), nominal (

C_{1}

), extended (

C_{2}

), shared/collision-risk (

C_{3}

), and joint-limit/singularity (

C_{4}

). A dedicated simulation environment built on ROS 2 Humble Hawksbill and Gazebo Classic 11 was used to generate a labelled dataset of 54,309 frames and 162,927 RGB images from three calibrated overhead cameras, with analytic ground-truth labels derived from closed-form forward kinematics. Training on a CPU with a feature-caching strategy brings the per-epoch wall-clock time to seconds, making the approach tractable without GPU hardware. On the held-out test set, the model achieves 87.1% overall accuracy and a macro-averaged F1 score of 90.0%, with near-perfect recall of 99.3% for the safety-critical shared zone

C_{3}

. The trained classifier is integrated as an ROS 2 inference node capable of running at 10

Hz

on a standard workstation. Our results demonstrate that joint angle information is a decisive complement to RGB imagery for fine-grained, safety-oriented workspace classification in simulation-derived settings. Full article

(This article belongs to the Topic Advances in Robot Vision Perception and Control Technology)

► Show Figures

Graphical abstract

17 pages, 2431 KB

Open AccessArticle

Bilevel Trajectory Optimization for Vacuum-Based Pick-and-Place Operations: A Numerical Study

by Georg Steinert, Clemens Troll and Jens-Peter Majschak

Robotics 2026, 15(7), 126; https://doi.org/10.3390/robotics15070126 - 30 Jun 2026

Viewed by 373

Abstract

Motion optimization has a significant influence on performance and robustness of modern robotic handling systems. In this study, a pick-and-place operation, as can be found in many processing machines, serves as a representative use case to develop a novel method for motion optimization. [...] Read more.

Motion optimization has a significant influence on performance and robustness of modern robotic handling systems. In this study, a pick-and-place operation, as can be found in many processing machines, serves as a representative use case to develop a novel method for motion optimization. Based on optimal control theory, the introduced method uses bilevel optimization simultaneously addressing process stability, favorable dynamic behavior and practical applicability. A process model for gripper load estimation established in the literature serves both as a basis for optimization and for evaluating the solution found. To establish a benchmark, a spline-based trajectory is generated. As this work proposes a theoretical approach, simulations based on the retrieved model serve as an evaluation basis of the resulting trajectories. The results show that a significant reduction in gripper load by approximately 60% was achieved compared to the reference motions. Eventually, requirements and limitations for application of the new method are discussed. Full article

(This article belongs to the Section Industrial Robots and Automation)

► Show Figures

Figure 1

30 pages, 9839 KB

Open AccessArticle

A Learning Framework for Robust Navigation of Mobile Robots Under Partial Observability

by Truong Nhut Huynh, Caiden Sivak, Hector Gutierrez and Kim-Doang Nguyen

Robotics 2026, 15(7), 125; https://doi.org/10.3390/robotics15070125 - 30 Jun 2026

Cited by 1 | Viewed by 520

Abstract

Autonomous navigation in mobile robotics faces tremendous challenges from partial observability due to sensor degradations such as noise and flickering in laser scans. Traditional methods like Adaptive Monte Carlo Localization (AMCL) and Gmapping perform well in ideal conditions but fail under these sensor [...] Read more.

Autonomous navigation in mobile robotics faces tremendous challenges from partial observability due to sensor degradations such as noise and flickering in laser scans. Traditional methods like Adaptive Monte Carlo Localization (AMCL) and Gmapping perform well in ideal conditions but fail under these sensor degradations. This paper develops a unified framework that integrates reinforcement learning with temporal sequence modeling, augmented by high-level semantic reasoning and parameterized quantum representations within a coherent architecture, to enable robust navigation for mobile robots. The framework models navigation as a partially observable Markov decision process (POMDP) and analyzes degraded LiDAR scans and odometry to generate velocity commands for motion planning and mapping. Experiments in a sim-to-real platform across four environments and real-world tests in indoor offices, outdoor terrains, and dynamic parking lots demonstrate substantial improvements compared to state-of-the-art methods. Success rates increase by up to 45 percentage points in dynamic scenarios, path lengths shorten by 20–25%, and map accuracies improve by 40% compared to baselines. The proposed approach achieves these gains through quantum-enhanced feature extraction for exploration, temporal modeling for state correction, and semantic reasoning for obstacle interpretation. This work advances reliable robot autonomy in uncertain environments. Full article

(This article belongs to the Section Sensors and Control in Robotics)

► Show Figures

Figure 1

18 pages, 13772 KB

Open AccessArticle

IPAnema Safe: A Cable-Driven Parallel Robot for Safe Operation Above People

by Christoph Martin, Marc Fabritius, Christian Lehnertz, Jakob Traub, Johannes T. Stoll, Werner Kraus and Andreas Pott

Robotics 2026, 15(7), 124; https://doi.org/10.3390/robotics15070124 - 30 Jun 2026

Viewed by 392

Abstract

Training ship crews for new scenarios at sea in reproducible sea conditions is weather-dependent and dangerous. The Maritime Research Institute Netherlands (MARIN) addresses this issue by offering a variety of different simulators to safely train people onshore. For scenarios involving a swinging hook [...] Read more.

Training ship crews for new scenarios at sea in reproducible sea conditions is weather-dependent and dangerous. The Maritime Research Institute Netherlands (MARIN) addresses this issue by offering a variety of different simulators to safely train people onshore. For scenarios involving a swinging hook or a ladder for a pilot boarding a ship in turbulent sea conditions, a safe cable-driven parallel robot, the IPAnema Safe, is developed. Its safety features allow its platform to move above people during training scenarios. Therefore, in addition to standard safety features and a proper mechanical design, the safety functions Safely Limited Position (SLP) and Safely Limited Speed (SLS) for a cable-driven parallel robot are implemented on safety-certified hardware with Performance Level e (PL e). This work presents the design process and implementation of the IPAnema Safe, focusing on its layout, winches, platform optimization and functional safety features. Full article

(This article belongs to the Section Intelligent Robots and Mechatronics)

► Show Figures

Figure 1

21 pages, 9002 KB

Open AccessSystematic Review

ROS-Enabled DIY and Open-Source Wheeled Robots for Higher Education Learning and Competitions: A Systematic Review

by Rúben Pereira, Benedita Malheiro and Manuel F. Silva

Robotics 2026, 15(7), 123; https://doi.org/10.3390/robotics15070123 - 30 Jun 2026

Viewed by 528

Abstract

This study systematically characterizes Do It Yourself (DIY) and open-source wheeled robotic platforms used in higher education and academic competitions. It also analyzes Robot Operating System (ROS)-based designs with respect to real-time performance and multi-sensor integration, following Preferred Reporting Items for Systematic Reviews [...] Read more.

This study systematically characterizes Do It Yourself (DIY) and open-source wheeled robotic platforms used in higher education and academic competitions. It also analyzes Robot Operating System (ROS)-based designs with respect to real-time performance and multi-sensor integration, following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. A total of 20 high-quality studies were identified across five major digital libraries (Dimensions, Web of Science, SpringerLink, ScienceDirect, and IEEE Xplore), which were searched on 12 January 2026. Eligibility was restricted to peer-reviewed English-language studies published between 2005 and 2026 that explicitly implement ROS-based wheeled platforms in higher education contexts. Results were synthesized through qualitative analysis using a structured data extraction form implemented in the Parsifal systematic review platform. Methodological quality and risk of bias were assessed using a structured appraisal checklist. The results show a dominant trend toward distributed dual-processor architectures, which separate low-level real-time control from high-level processing. Most platforms target an accessible price range of 50€ to 500€ for open-source and DIY platforms. ROS has emerged as the standard middleware, enabling multi-sensor integration and supporting digital twin workflows. There is also a clear shift toward open-source hardware and Three-Dimensional (3D)-printed modular designs, which reduce production costs. However, challenges remain, including software obsolescence and the lack of maintenance plans. The findings highlight the need for interoperable reference architectures and automated deployment workflows to ensure long-term sustainability. Evidence is limited by heterogeneity, inconsistent reporting, and small sample sizes, which introduce risks of bias and imprecision. This review was formally registered with protocols.io. Full article

(This article belongs to the Section Educational Robotics)

► Show Figures

Figure 1

33 pages, 55733 KB

Open AccessArticle

LE-HG-PRM: A Structure-Aware Roadmap Planner for Intelligent Warehouse Logistics

by Siyuan Wang, Gongsen Wang, Feng Yang, Dawu Peng, Xingyu Yan, Shuyi Zhang, Xinyi Li and Zhen Tian

Robotics 2026, 15(7), 122; https://doi.org/10.3390/robotics15070122 - 29 Jun 2026

Viewed by 377

Abstract

Efficient AGV/AMR path planning is essential for intelligent warehouse logistics, where regular shelves, narrow aisles, local bottlenecks, and heterogeneous obstacles strongly affect roadmap quality. This study proposes LE-HG-PRM, a structure-aware extension of heuristic-guided probabilistic roadmap planning. The method embeds warehouse geometric priors into [...] Read more.

Efficient AGV/AMR path planning is essential for intelligent warehouse logistics, where regular shelves, narrow aisles, local bottlenecks, and heterogeneous obstacles strongly affect roadmap quality. This study proposes LE-HG-PRM, a structure-aware extension of heuristic-guided probabilistic roadmap planning. The method embeds warehouse geometric priors into probability-field sampling, region-adaptive neighborhood connection, and cache-accelerated progressive path refinement. Compared with the preliminary conference version, the journal version introduces a redesigned warehouse-oriented planning framework and substantially expands the experimental validation. Four experimental campaigns are conducted, covering static-complexity progression, corridor-width sensitivity, parameter sensitivity, and map-scale expansion, with A*, JPS, PRM, RRT, RRT*, and HG-PRM as baselines. Each scenario uses 50 paired start–goal tasks, and sampling-based methods are repeated with 12 independent random seeds. The results show that LE-HG-PRM provides competitive path quality and structurally regular paths in representative warehouse layouts. Statistical tests further confirm that its path-length advantage is scenario-dependent but significant in several structured and bottleneck-constrained settings. The findings suggest that incorporating explicit warehouse-structure priors can improve roadmap-based global planning for intelligent logistics, while future work should validate the method in Gazebo and physical AGV/AMR platforms. Full article

(This article belongs to the Special Issue Embodied AI for Soft and Bio-Inspired Robotics)

► Show Figures

Figure 1

20 pages, 331 KB

Open AccessReview

Nonverbal Auditory Communication for Human–Robot Interaction in Industry 5.0: A Scoping Review

by Tom Schmid, Manja Lohse, Sven Winkelmann and Alexander von Hoffmann

Robotics 2026, 15(7), 121; https://doi.org/10.3390/robotics15070121 - 26 Jun 2026

Viewed by 739

Abstract

In Industry 5.0 (I5.0), close-proximity human–robot collaboration demands communication beyond conventional alarms and speech. Nonverbal auditory communication offers a complementary modality, yet its role in I5.0 remains unmapped. This scoping review maps nonverbal auditory communication research in I5.0 Human–Robot Interaction (HRI) and compares [...] Read more.

In Industry 5.0 (I5.0), close-proximity human–robot collaboration demands communication beyond conventional alarms and speech. Nonverbal auditory communication offers a complementary modality, yet its role in I5.0 remains unmapped. This scoping review maps nonverbal auditory communication research in I5.0 Human–Robot Interaction (HRI) and compares it with general HRI literature to identify transfer potential and research gaps. Peer-reviewed English-language articles (2023–April 2026) addressing nonverbal sound in HRI contexts were included. Speech, emotion detection, haptic interfaces and non-HRI domains were excluded. A search with two syntaxes across Web of Science, Scopus, IEEE Xplore, ACM and MDPI, supplemented by citation searching, targeted I5.0-specific (Syntax S1) and general HRI auditory literature (Syntax S2). This created two article record sets, n1 and n2. Articles were organized following Arksey and O’Malley’s framework and PRISMA-ScR into four inductively derived clusters: Sonification, Multimodal Feedback Systems, Safety and Frameworks and Concepts. From 782 initial records, 16 (n1) and 32 (n2) articles were included. In I5.0, multimodal feedback dominates: intentionally designed nonverbal sounds improve situational awareness, reduce cognitive workload and increase perceived safety. Compared to n2, which is shaped by social robotics and emotion-driven sound design, five gaps emerge in I5.0: absent emotion-related sound perception research, missing field studies, missing industry-specific sound design frameworks, underutilized sonification for spatial awareness and safety and no unimodal auditory studies under realistic industrial conditions. A dedicated sound design framework operationalizing I5.0 communicative requirements into designable sound parameters is needed, alongside empirical validation under realistic industrial noise conditions. Full article

(This article belongs to the Special Issue Human–Robot Collaboration in Industry 5.0)

► Show Figures

Figure 1

26 pages, 2428 KB

Open AccessArticle

Reconfigurable Mobile Wireless Sensor Network Coordination for Simultaneous Multi-Target Tracking

by Naeimeh Najafizadeh Sari, Yeqi Sang, Goldie Nejat and Beno Benhabib

Robotics 2026, 15(7), 120; https://doi.org/10.3390/robotics15070120 - 25 Jun 2026

Viewed by 629

Abstract

This paper presents a distributed coordination framework for simultaneous multi-target tracking using a mobile wireless sensor network (MWSN) based on discrete-event-system principles. The proposed framework employs a finite-state-machine architecture, where autonomous mobile sensors sequentially process detection and tracking events. Unlike passive tracking approaches [...] Read more.

This paper presents a distributed coordination framework for simultaneous multi-target tracking using a mobile wireless sensor network (MWSN) based on discrete-event-system principles. The proposed framework employs a finite-state-machine architecture, where autonomous mobile sensors sequentially process detection and tracking events. Unlike passive tracking approaches that react to target loss after it occurs, the proposed strategy implements predictive handover through Extended-Kalman-Filter-based uncertainty propagation. This enables sensors to anticipate target loss and to reposition auxiliary sensors in advance, acquiring targets along their predicted trajectories. A bidding-based allocation mechanism coordinates sensor assignments by evaluating four competing objectives: network preservation, spatial proximity to handover points, temporal mission feasibility, and estimation uncertainty. The proposed framework integrates four components: EKF-convergence-triggered proactive handover, multi-objective competitive bidding, distributed min–max conflict resolution, and fusion-driven proportional navigation. Unlike existing methods, auxiliary sensors navigate using confidence-weighted EKF estimates shared by neighboring sensors rather than their own measurements. An ablation study over ten Monte Carlo trials confirms that each component contributes independently, with EKF-based predictive triggering identified as the dominant performance driver. Full article

(This article belongs to the Section Sensors and Control in Robotics)

► Show Figures

Graphical abstract

37 pages, 2675 KB

Open AccessArticle

Decentralized Shared Actor–Critic Learning for Collision-Aware Small-Team Multi-Robot Coverage

by Abzal E. Kyzyrkanov, Didar Yedilkhan, Saltanat Amirgaliyeva and Sergazy Narynov

Robotics 2026, 15(7), 119; https://doi.org/10.3390/robotics15070119 - 25 Jun 2026

Viewed by 610

Abstract

This study presents a decentralized shared actor–critic framework for cooperative multi-robot coverage in continuous two-dimensional simulation. The method combines permutation-invariant local observations, continuous differential-drive control, and reward shaping based on stepwise Hungarian assignment distances, collision penalties, and time efficiency. Homogeneous teams of four, [...] Read more.

This study presents a decentralized shared actor–critic framework for cooperative multi-robot coverage in continuous two-dimensional simulation. The method combines permutation-invariant local observations, continuous differential-drive control, and reward shaping based on stepwise Hungarian assignment distances, collision penalties, and time efficiency. Homogeneous teams of four, five, and six agents are evaluated in an obstacle-free environment using five independent training seeds. In the final training window, the full reward configuration achieved full-team success rates of 98.2 ± 2.9% for four agents, 85.1 ± 18.0% for five agents, and 96.3 ± 2.0% for six agents, with mean landmark coverage above 96% in all cases. The lower mean in the five-agent setting was associated with higher seed-level variability dominated by one low-success seed. Reward ablations without assignment shaping or collision penalties remained viable, and seed-level tests did not show a statistically significant final-window advantage of the full reward configuration. The full configuration reached the 80% rolling-success threshold earlier in median terms, with the clearest seed-level support in the four-agent setting. Within-environment comparison showed higher full-team success than MADDPG and MAPPO under the matched training horizon and final-window protocol. Deterministic arena-size transfer from

15 \times 15

to

30 \times 30

showed decreasing full-team success as arena size increased, while partial landmark coverage remained higher than strict full-team completion. The results support the method for small homogeneous teams in the tested obstacle-free simulation, while larger teams, external obstacles, aerial-robot dynamics, formal safety guarantees, and hardware deployment remain future work. Full article

(This article belongs to the Special Issue AI-Powered Robotic Systems: Learning, Perception and Decision-Making)

► Show Figures

Figure 1

17 pages, 15730 KB

Open AccessArticle

Gynecological Tendon-Driven Continuum Robots: Design and Experimentation

by Clara G. Kierbel and Matteo Russo

Robotics 2026, 15(7), 118; https://doi.org/10.3390/robotics15070118 - 25 Jun 2026

Viewed by 544

Abstract

Most gynecological interventions do not take advantage of the possible access through the natural orifice to the operating zone and/or use rigid tools, which leads to more invasive procedures. The purpose of this research is to reduce invasiveness by creating a natural orifice [...] Read more.

Most gynecological interventions do not take advantage of the possible access through the natural orifice to the operating zone and/or use rigid tools, which leads to more invasive procedures. The purpose of this research is to reduce invasiveness by creating a natural orifice endoscopic surgical tool. By analyzing the varied anatomies present in patients to extract functional requirements, we propose a conceptual design that allows for better navigation of the environment thanks to a custom design with active control over endoscope shape. We manufactured and tested this new design of a tendon-driven continuum robot in a phantom that is representative of the geometrical properties and variability of a uterus, validating its operation and functionality. Full article

(This article belongs to the Special Issue Selected Papers from MESROB2025: 9th IFToMM International Workshop on New Trends in Medical and Service Robotics)

► Show Figures

Figure 1

5 pages, 154 KB

Open AccessEditorial

Applications of Neural Networks in Robot Control

by Luca Patanè and Paolo Arena

Robotics 2026, 15(7), 117; https://doi.org/10.3390/robotics15070117 - 23 Jun 2026

Viewed by 545

Abstract

This Editorial introduces the Special Issue “Neural Networks for Robot Control”, which gathers contributions that reflect the rapidly growing intersection of machine learning and robotics [...] Full article

(This article belongs to the Special Issue Applications of Neural Networks in Robot Control)

Journal Menu

Journal Browser

Robotics, Volume 15, Issue 7 (July 2026) – 23 articles

Further Information

Guidelines

MDPI Initiatives

Follow MDPI