Runjie Shen

Robots operating in daily environments, such as householdorganization and object manipu lation, often encounter densely cluttered spaces where collision-free motion is infeasible. To safely and effectively interact with surrounding objects, robots require task-agnostic whole body contact-rich motion planning that can handle time-varying multi-contact interactions. However, model-based approaches are limited by the difficulty of accurately estimating unknown interaction dynamics online, while learning-based approaches typically rely on task-specific demonstrations or reward functions, restricting their generalization to unseen tasks.This paper proposes a physics-model-guided whole-body contact-rich motion planning framework based on material-point morphology representation and interaction motion prediction supervision. The robot body is represented by a set of material points with consistent body correspondence, enabling unified encoding of whole-body motion and tactile interaction states under varying contact locations, contact configurations, and force distributions. Furthermore, interaction motion prediction generated from a tactile-driven interaction planner is used as physical supervision to train MorphoPlanner, an end-to-end motion morphology planning network that generates future whole-body motion references from task objectives and current interaction states without requiring task-specific demonstrations or manually designed rewards. A whole-body tactile interaction dataset containing 749K samples is constructed for model training and evaluation. Extensive experiments, including compliant interaction tests, stress tests in highly cluttered environments, and comparisons with representative planning methods, are conducted on simulation and real-world dual-arm tactile platforms. The proposed framework achieves success rates of 95.5%-100% in dense cluttered scenarios and demonstrates stable whole-body contact-rich motion generation under real-world physical interactions. These results validate the effectiveness of the proposed morphology-aware planning framework for improving robot adaptability in unknown and cluttered environments.

The orchestration of heterogeneous swarm systems, such as Unmanned Aerial Vehicle (UAV) and Unmanned Ground Vehicle (UGV) teams, in unknown, GPS-denied environments requires coordination among active perception, collision avoidance, and topological flexibility. Traditional Artificial Potential Field (APF) methods can suffer from local-minimum deadlocks, while optimization-based Control Barrier Function Quadratic Programs (CBF-QPs) may produce high-frequency control variation in narrow corridors. To address these limitations, this paper proposes the Riemannian Information-geometric Swarm Motion Orchestration (RISMO) framework. RISMO integrates objective-driven navigation, perception-aware motion generation, and obstacle avoidance within a common non-Euclidean metric representation. We construct a Riemannian metric tensor that incorporates the Fisher Information Matrix (FIM) to encourage the UGVs to preserve line-of-sight tracking of the UAVs. To accommodate formation deformation in narrow passages, the swarm topology is modeled as a Semi-Markov Jump System (S-MJS). Under the enforced positive per-agent inter-event bound and the stated Average Dwell Time (ADT) condition, the modeled switching sequence has a bounded transition frequency and cannot exhibit Zeno accumulation. Furthermore, a goal-aligned tangential flux supplies a deterministic local directional bias near saddle-like stagnation. The numerical and ablation results show that this mechanism facilitates escape from the tested non-convex traps. In the planar simulations, RISMO reduces the observed gradient stagnation of APF and the high-frequency control variation of CBF-QP while preserving positive obstacle clearance and cooperative perception.

LiDAR-based Simultaneous Localization and Mapping (SLAM) systems suffer from environment-induced degeneracy in geometrically self-similar scenes such as tunnels and cylindrical structures, where the point cloud registration problem becomes ill-conditioned. Existing degeneracy detection methods predominantly rely on eigenvalue analysis of the Hessian matrix, which assesses geometric conditioning at a fixed linearization point. However, in semi-degenerate environments containing partial but sufficient geometric constraints, these methods tend to produce false positive detections because measurement noise inflates eigenvalues and obscures the contribution of minor structural features. In this paper, we propose the Optimization Convergence Efficiency (OCE) metric, which evaluates the relative contraction of the directional optimizer gradient across Levenberg-Marquardt evaluations and directly characterizes the convergence response of LiDAR registration along the monitored state direction. The proposed method is evaluated through controlled simulation with ground-truth-based quantitative assessment and real-world UAV experiments in a wind-turbine tower and a long tunnel. In the nominal simulation, OCE achieves 98.21% recall with a 2.33% false-positive rate, compared with 100% recall and a 49.42% false-positive rate for the conventional eigenvalue-ratio detector. Controlled measurement-noise experiments further demonstrate that OCE maintains substantially lower false-alarm rates as spectral discrimination becomes less reliable. These results show that OCE can retain usable LiDAR pose estimates in semi-degenerate environments while reliably identifying actual registration failure.

Recent autonomous exploration systems for unmanned platforms have improved motion efficiency, trajectory smoothness, and replanning continuity, but thorough coverage remains difficult in confined environments with short branches and partially resolved connectors. We present TOVEX, a completion-oriented framework that couples unresolved Euclidean signed distance field (ESDF) voxels with a sparse topological voxel-sphere (TVS) graph. Evaluation includes a bridge simulation, five-run maze comparisons with the baseline systems EDEN, GBPlanner, and TARE, component ablation, runtime and memory profiling, a cross-structure cave-scene stress test, and real-world experiments with an unmanned aerial vehicle (UAV) in an underground garage and an unmanned ground vehicle (UGV) in a warehouse. The bridge test shows coherent mapping and continuous velocity profiles. In the maze, TOVEX achieves the highest median final coverage of 96.2%, exceeding TARE by 5.2 percentage points, while TARE remains stronger in early coverage and motion efficiency. The geometry-derived terminal-tail analysis confirms that TOVEX achieves the most complete residual-passage resolution, while removing short-branch priority produces the largest loss. At approximately 1100 TVS vertices, median target-selection latency is 0.89 ms and planner proportional set size (PSS) is 52.3 MiB. Field trials demonstrate the same gain-bearing decision layer on aerial and ground platforms. TOVEX is well suited to unmanned-platform missions that prioritize resolving accessible residual passages.

In GNSS-denied pipeline inspection (low light, confined space, repetitive geometry), LiDAR–inertial odometry (LIO) on UAVs often suffers from poor observability in axial translation and rotation about the pipe axis, leading to rapid drift and potential collisions. We propose an intensity-aided, degeneracy-aware localization framework. First, we perform an online decomposed observability analysis by inspecting covariance eigenvalues to identify degenerate translation and rotation directions. Second, we construct a 2D representation of LiDAR intensity and extract intensity feature blocks that provide complementary constraints when geometric structure is insufficient. Finally, geometric and intensity residuals are tightly fused within an iterated error-state Kalman filter (IESKF) to stabilize pose estimation under severe degeneracy. Experiments in ROS/Gazebo simulation and real pipeline data demonstrate improved robustness, with reduced axial drift and more accurate yaw estimation compared with representative baselines.

In this paper, we present the design and control of a novel aerial vehicle inspired by the biomechanics of a rabbit named “Flybbit”. Flybbit consists of two main components, namely a movable “Ears” part and a rigid “Body” part, forming a composite flying system with five controllable degrees of freedom (DOFs). The “Ears” part is equipped with two tiltable motors paired with optional-sized propellers, enabling additional thrust generation and flight stability maintenance, and the “Body” part incorporates four fixed motors, analogous to a rabbit’s limbs, to provide the primary propulsion. To fully exploit the actuation capability, we derive the system dynamics and introduce a dynamic control allocation method with an adaptive strategy to mitigate actuator saturation during complex combined maneuvers. Furthermore, we analyze the differential flatness property and develop a nonlinear inverse dynamics controller enhanced with hybrid external wrench estimation, enabling accurate trajectory tracking in five DOFs. Flybbit supports both manual operation via RC and autonomous flight via onboard computation. Comprehensive simulations and real-world experiments validate the proposed design and control framework.

Unmanned aerial vehicle (UAV) navigation systems face significant challenges in complex environments, such as sensor degradation and signal loss. This study proposes the NSDDI-AFF (Normalized Single-Dimensional Degradation Isolation with Adaptive Federated Filtering) method to address these issues. By integrating adaptive thresholding, degradation isolation, and multi-sensor fusion, the method dynamically identifies degraded channels and ensures robust state estimation. Evaluated in a semi-enclosed coal yard and in a hydropower pipeline, NSDDI-AFF reduced positional errors by 97.5% and 95.7% compared to LIO-SAM and Fast-LIO2, achieving a position RMSE of 0.05 m and orientation RMSE of 0.5°. The method detected degradation early (120 s) and maintained mapping accuracy with geometric errors below 0.5%. These results demonstrate that NSDDI-AFF significantly enhances UAV positioning accuracy, fault tolerance, and mapping reliability, making it a robust solution for challenging industrial applications.

Simultaneous Location and Mapping (SLAM) is a common algorithm for position estimation in GNSS-denied environments. However, the high structural consistency and low lighting conditions in tunnel environments pose challenges for traditional visual SLAM and LiDAR SLAM. To this end, this paper presents LiDAR and optical flow fusion odometry (LOFF), which uses a direction-separated data fusion method to fuse optical flow odometry into the degenerate direction of the LiDAR SLAM without sacrificing the accuracy. Moreover, LOFF incorporates detectors and a compensator, allowing for a smooth transition between general environments and degeneracy environments. This capability facilitates the stable flight of unmanned aerial vehicles (UAVs) in GNSS-denied tunnel environments, including corners and long-distance consistency. Through real-world experiments conducted in a GNSS-denied pedestrian tunnel, we demonstrate the superior position accuracy and trajectory smoothness of LOFF compared to state-of-the-art visual SLAM and LiDAR SLAM.