collision avoidance

This paper addresses collaborative motion planning for homogeneous linear multi-agent systems operating in unknown obstacle-rich environments without explicit system models. Method: We propose a fully data-driven framework that is dynamically feasible and provably safe. It learns feedback gains and local invariant ellipsoids—serving as safety certificates—by solving a semidefinite program on experimental data. Distributed, optimization-free trajectory generation is achieved by integrating grid-based RRT sampling with a spatiotemporal resource reservation mechanism. Contribution/Results: To the best of our knowledge, this is the first work to unify data-driven invariant set learning with spatiotemporal reservation. Relying solely on limited experimental data and convex optimization tools, it simultaneously guarantees collision avoidance with static/dynamic obstacles and inter-agent collisions. The framework significantly reduces computational overhead while providing formal safety guarantees. Extensive simulations validate its effectiveness under tight dynamical constraints and complex obstacle configurations.

Multi-agent systems under limited communication range lack formal infinite-horizon safety guarantees. Method: This paper proposes the first decentralized asynchronous motion planning framework that provides rigorous collision-avoidance safety for nonlinear Dubins-type agents operating over dynamic topologies. It integrates Guardian-based safety control with R-bounded geometric constraints to establish, for the first time, a theoretical linkage between communication radius and distributed safety planning capability. Forward invariance analysis and locally informed asynchronous trajectory optimization ensure infinite-horizon safety using only neighbor-to-neighbor communication. Results: In high-density simulations with 128 agents, the framework achieves 100% collision-free operation, with safety performance invariant to system scale—demonstrating both formally provable safety and computationally scalable decentralized planning.

This work addresses real-time trajectory optimization and cooperative control of autonomous agents on resource-constrained edge devices. We propose an efficient Model Predictive Control (MPC) framework based on integral Chebyshev collocation. Our key contribution is the first integration of integral Chebyshev polynomial parameterization with differentiable polyhedral collision checking, enabling explicit modeling of actuator saturation and hard obstacle-avoidance constraints. The formulation minimizes L₂ approximation error and is solved via quadratic programming for rapid convergence. Employing a receding-horizon MPC architecture, the method achieves over 3.2× speedup over conventional approaches on edge hardware, enabling sub-millisecond replanning. We validate its safety, real-time performance, and cooperative capability in multi-spacecraft formation control tasks, demonstrating robust constraint satisfaction and scalable coordination under tight computational budgets.

To address deadlock and collision issues among multiple robots navigating constrained environments—such as doorways and intersections—in confined spaces, this paper proposes a decentralized navigation framework. The system is modeled as a non-cooperative, communication-free, partially observable “social micro-game,” and introduces, for the first time, a minimally intrusive liveness guarantee mechanism based on pre-perturbed barrier certificates (PBCs), ensuring each robot autonomously transitions into a deadlock-free state set. The approach integrates discrete-time control barrier functions (DCBFs), receding-horizon optimization, and distributed real-time feedback control. Experimental validation on F1/10, Jackal, and Spot platforms demonstrates a 23% improvement in task success rate, zero collisions, a 41% reduction in average stopping time, and significantly enhanced path smoothness and throughput compared to both state-of-the-art decentralized and centralized methods.

This work addresses safe collaborative navigation for multi-robot systems without individual reference trajectories. Methodologically, it proposes a behavior-driven safe multi-agent reinforcement learning framework that employs only the formation centroid as the navigation target—eliminating conventional per-robot path planners—and integrates model predictive control (MPC) as an online safety filter to explicitly guarantee collision-free operation during both training and deployment. To our knowledge, this is the first approach achieving provably safe collaborative navigation under the no-individual-reference setting. The MPC constraints not only accelerate policy convergence but also enable safe online deployment on real robots even in early training stages. Extensive simulations and real-world experiments demonstrate zero collisions, faster target arrival compared to baselines, and robust practical performance—validating both efficacy and deployability.

This study addresses the challenge of cooperative collision avoidance among multiple spacecraft under intermittent ground station communication constraints. The authors propose a semi-decentralized partially observable Markov decision process (SDec-POMDP) framework that explicitly incorporates ground station visibility into the multi-agent decision-making model for the first time. To solve for joint maneuver strategies, they design an approximate recursive short-horizon semi-decentralized A* algorithm (RS-SDA*). Operating solely within actual communication windows, this approach significantly reduces coordination synchronization events—by 28.5% compared to continuous coordination—while closely approximating the maneuver performance of centralized planning. Moreover, it satisfies safety distance constraints more consistently than heuristic rule-based methods and minimizes unnecessary orbital deviations.

This work proposes an online trajectory generation method based on piecewise quintic/quartic splines to address the challenge of converting arbitrary geometric paths into kinematically feasible and collision-free trajectories in dynamic environments. The approach explicitly enforces jerk constraints and supports real-time replanning under high-frequency goal updates. By integrating dynamic environment perception and a responsive adaptation mechanism, it guarantees collision avoidance within finite time while permitting bounded deviations from the original path. Both simulation and real-world experiments demonstrate that the method outperforms existing approaches in trajectory smoothness, computational efficiency, and real-time performance, achieving stable operation in human-in-the-loop dynamic scenarios with target update rates up to 1 kHz.

This work addresses the challenges of decentralized collision avoidance for nonlinear multi-agent systems in the absence of trajectory information exchange, where existing approaches often suffer from excessive conservativeness and difficulties in guaranteeing recursive feasibility and convergence. The paper proposes a safety-set-based decentralized emergency model predictive control framework, wherein each agent relies solely on its local state and employs a consensus rule to couple nominal trajectories with emergency certificates to ensure collision avoidance. A novel geometric safety set update mechanism is introduced to guarantee recursive feasibility across consecutive time steps, complemented by a Lyapunov-like convergence theory that enables plug-and-play operation. The method demonstrates effective, robust, and scalable collision-free navigation in both sparse and dense environments, including those with complex bottlenecks.

In dynamic and uncertain environments, ensuring safety, robustness, and scalability simultaneously in multi-agent motion planning remains challenging. To address this, we propose RE-DPG—a decentralized cooperative decision-making framework integrating dynamic potential games with multi-agent forward reachable sets (MA-FRS) under local interactions. We introduce two novel algorithms: neighborhood-dominant iterative best response (ND-iBR) and iterative ε-best response (iε-BR), enabling rapid convergence to an ε-Nash equilibrium with explicit safety constraints. Theoretical analysis guarantees convergence and establishes proactive safety margins. Evaluated in 2D/3D simulations and on real robotic platforms, RE-DPG significantly improves planning efficiency and obstacle-avoidance reliability while scaling to hundreds of agents. It achieves strong robustness against environmental uncertainty and computational scalability through distributed computation.

Addressing the lack of formal safety guarantees and dynamic adaptability for multi-vehicle formation flight in complex 3D urban air mobility (UAM) environments, this paper proposes a safety-critical cooperative framework based on a leader–follower architecture. The method integrates a precomputed backup trajectory set, an online formation tracking controller, and leader-path-dependent safe backup maneuvers—extending the Guardian algorithm to 3D multi-robot systems for the first time. All components are formally verified to ensure collision avoidance under bounded disturbances and actuation limits. The framework achieves provable safety without sacrificing real-time performance: in 100 randomized high-fidelity simulations, it attains 100% obstacle avoidance success, outperforming both Control Barrier Function (CBF) and Nonlinear Model Predictive Control (NMPC) baselines. Furthermore, experimental validation is conducted on a physical quadcopter swarm, confirming practical feasibility and robustness in real-world deployment.