A Distributed Algorithm for Real-Time Multi-Drone Collision-Free Trajectory Replanning

Abstract

In this paper, we present a distributed algorithm to generate collision-free trajectories for a group of quadrotors flying through a common workspace. In the setup adopted, each vehicle replans its trajectory, in a receding horizon manner, by solving a small-scale optimization problem that only involves its own individual variables. We adopt the Voronoi partitioning of space to derive local constraints that guarantee collision avoidance with all neighbors for a certain time horizon. The obtained set of collision avoidance constraints explicitly takes into account the vehicle’s orientation to avoid infeasiblity issues caused by ignoring the quadrotor’s rotational motion. Moreover, the resulting constraints can be expressed as Bézier curves, and thus can be evaluated efficiently, without discretization, to ensure that collision avoidance requirements are satisfied at any time instant, even for an extended planning horizon. The proposed approach is validated through extensive simulations with up to 100 drones. The results show that the proposed method has a higher success rate at finding collision-free trajectories for large groups of drones compared to other Voronoi diagram-based methods.

A Self-Collision Detection Algorithm of a Dual-Manipulator System Based on GJK and Deep Learning

“…The GJK algorithm generates a simplex in every iteration process, and the simplex generated at the k -th iteration process is denoted as . is the point nearest to the origin in and can be calculated by choosing the Johnson operator [ 21 ] or the improved operator [ 28 , 35 , 36 ] depending on the situation. If is the origin, then contains the origin, which means that contains the origin, and thus, a collision occurs between and .…”

“…Those of [ 27 ] applied the GJK algorithm to the field of robotics to optimize the gripping force of the robotic arm on the target object, via the accurate calculation of the distance between convex objects. The authors of [ 28 ] used the GJK algorithm for collision detection in Unmanned Aerial Vehicle (UAV) swarm trajectory planning and improved the distance operator by combining the usage scenarios. After simulation and actual robot validation, the GJK algorithm of the original authors was selected for self-collision detection in our research.…”

Synchronous Control of a Group of Flying Robots Following a Leader UAV in an Unfamiliar Environment

“…Another approach is the simultaneous monitoring and remote control of each vehicle in the group [ 4 ]. Theoretically, it is possible to plan a trajectory for multiple drones in a constraint area [ 5 ]. However, in a natural environment, flights are only collision-free for a short time, because of unexpected disturbances.…”

Robust MADER: Decentralized and Asynchronous Multiagent Trajectory Planner Robust to Communication Delay

“…Centralized planners [8], [9] have one entity that directs all agent trajectories, providing simplicity but with the drawback of a potential single point of failure. Conversely, decentralized planners [4]- [7], [12], [14] enable each agent to plan its own trajectory, offering better scalability and resilience. Similarly synchronous planners [7], [11], [26] require synchronization before planning, whereas asynchronous methods [4]- [6] let agents plan individually, proving to be more scalable.…”