The reliability of quadcopters is becoming increasingly important. However, the field of quadcopter fault diagnosis lacks a standardized, reproducible benchmark for fair and comprehensive algorithm evaluation. To address this gap, this paper presents RflyBench-FD, a standardized, open benchmark for quadcopter fault diagnosis built upon the RflyMAD dataset, publicly accessible at https://www.rflybench.com/competitions/5/ . RflyBench-FD establishes a complete evaluation framework that covers eight representative fault types and provides standardized simulation/real dual-domain splits, a well-defined task formulated as eight parallel classification sub-tasks, publicly released scoring scripts, and a multi-dimensional metric system (accuracy, F1-score, and robustness under Gaussian noise at 8/4/2 dB SNR and physically motivated sensor degradations) aggregated into a weighted composite score whose ranking stability is verified by a sensitivity analysis. To validate the benchmark, we evaluate twelve baselines spanning classical machine learning, temporal deep learning, domain adaptation, transfer learning, and training-mode ablations, complemented by noise tests, physically motivated sensor-degradation tests, leakage-oriented checks, and paired significance tests; the results show that the benchmark produces stable and discriminative rankings across method families. The platform further supports daily submissions, a governed real-time leaderboard with anti-overfitting safeguards, full result traceability, and an LLMassisted analysis report. By providing a fair, reproducible, and insight-rich evaluation infrastructure, RflyBench-FD aims to support the development and real-world deployment of robust fault diagnosis solutions for autonomous quadcopters.
This paper investigates a state-based safe trajectory tracking (STT) control framework for autonomous vehicles in dynamic environments, where the objective is to design a controller that ensures accurate trajectory tracking while guaranteeing non-conservative collision avoidance. Existing learning-based approaches often suffer from inaccurate geometric safety representations and unstable training when incorporating safety constraints. To address these challenges, we propose a geometry-aware framework by integrating an analytical rotating supporting hyperplane (RSH)-based safety representation with a Lagrangian Proximal Policy Optimization (Lagrangian-PPO) reinforcement learning algorithm. By modeling vehicles and obstacles as rigid ellipsoids, an explicitly differentiable signed distance is derived to provide a continuous and non-conservative safety cost. Based on this representation, the trajectory tracking problem is formulated as a constrained Markov decision process, enabling unified optimization of tracking performance and safety via a reward–cost decoupled learning structure. Finally, asymptotic convergence and effectiveness of the proposed approach are validated through simulation and Vehicle-in-the-Loop (ViL) experiments, demonstrating superior tracking accuracy and robust obstacle avoidance compared with existing methods.
In surface-to-air interception, the initial and midcourse guidance phases of the interceptor have an important influence on the overall interception performance. To address this problem, this paper proposes a trajectory planning method for the initial and midcourse guidance phases based on the Sequential Quadratic Programming (SQP) optimization algorithm. For highly maneuvering targets, a receding horizon optimization strategy is introduced into the trajectory optimization process. The interception trajectory is updated in real time through multi-stage optimization. This improves the adaptability and robustness of the guidance planning method. In addition, terminal guidance planning is included in the overall trajectory planning framework. This provides favorable conditions for a smooth handover from midcourse guidance to terminal guidance. To track the planned trajectory accurately, a Model Compensation Control (MCC) method based on a Compensation Function Observer (CFO) is adopted. Finally, simulation experiments are conducted to verify the effectiveness and robustness of the proposed algorithm. Based on this method, the interceptable range of a single interceptor is also calculated.
Collaborative exploration by unmanned vehicles and drones relies on precise positioning and mapping, as well as efficient path planning. However, inconsistencies in spatial references among cross-platform sensors and the lack of environmental description due to sparse features lead to relative pose estimation drift during collaboration, making it difficult to support high-precision navigation for autonomous vehicles. To address the aforementioned issues, this paper proposes an integrated air-ground collaborative exploration method that integrates collaborative localization and mapping technology with an improved JPS path planning algorithm. At the perception level, we improved the ORB-SLAM2 algorithm by introducing a 3D Gaussian field as the underlying map representation and combined it with the G-ICP algorithm to solve the problem of cross-view point cloud registration, thus constructing a dense environment map with both global consistency and detailed texture. This high-precision global prior information effectively compensates for the blind spots in the local perception of autonomous vehicles, providing reliable prior environmental information for path planning. Based on this, to address the problems of traditional planning algorithms being prone to local optima and poor trajectory smoothness during the exploration process, we have integrated an improved JPS algorithm with multi-dimensional constraints: by introducing dynamic weight factors to balance heuristic estimation and actual cost, and by using direction factors and line-of-sight detection mechanisms to optimize the search strategy. This method significantly improves the smoothness of the trajectory’s compliance with the kinematic constraints of the autonomous vehicle while ensuring global optimality. Finally, complex scenario tests were conducted using a real heterogeneous robot collaborative platform. The results show that the success rate of collaborative exploration between the autonomous vehicle and the drone in complex environments reached 87.5%, validating the robustness and superiority of the proposed method in environmental map construction and path planning during collaborative exploration between drones and autonomous vehicles.
The equations of motion of a flying robot composed of n rigid arms, each equipped with a motor at its end to provide controllability during complex three-dimensional maneuvers, were previously derived in [1]. The most significant outcome of that study was the demonstration that the robot can reach any point in space while simultaneously achieving an arbitrary desired orientation. It was shown that accomplishing this objective requires at least six actuators and, consequently, six propellers. However, increasing the number of actuators results in higher manufacturing costs, greater energy consumption, and increased overall weight. Therefore, the development and analysis of more efficient flying robot configurations become increasingly important. In the design proposed in this paper, a three-arm flying base is employed instead of a six-arm configuration, with a propeller mounted at the end of each arm. A distinctive feature of this design is that each propeller, in addition to rotating about its primary axis, can also rotate about a secondary axis. This capability enables the thrust vector generated by the main propeller motor to be reoriented by means of a smaller auxiliary motor, thereby allowing the robot to achieve arbitrary orientations in space. To facilitate the analysis, the flying robot is decomposed into several open-chain robotic structures. The equations of motion of the individual chains are first derived using the Gibbs–Appell formulation and then combined, while accounting for the constraints governing the interactions among the chains, to obtain the dynamic model of the complete system. Finally, the capability of the proposed tiltrotor tricopter to perform complex three-dimensional maneuvers is evaluated through an integrated feedback-linearization and LQR–PD control framework.
