Trajectory Planning for Heterogeneous Robot Teams

Neural-Swarm2: Planning and Control of Heterogeneous Multirotor Swarms Using Learned Interactions

“…For motion planning, empirical models have been used to avoid harmful interactions [2,4,6,26,27]. Typical safe boundaries along multi-vehicle motions form ellipsoids [4] or cylinders [6] along the motion trajectories.…”

“…For motion planning, empirical models have been used to avoid harmful interactions [2,4,6,26,27]. Typical safe boundaries along multi-vehicle motions form ellipsoids [4] or cylinders [6] along the motion trajectories. Estimating such shapes experimentally would potentially lead to many collisions and dangerous flight tests and those collision-free regions are in general conservative.…”

“…where U (k) is the control space for type k robots, X (k) is the free space for type k robots, r (lk) is the minimum safety distance between type l and type k robots, f (k) a,max is the maximum endurable interaction force for type k robots, τ (k) a,max is the maximum endurable interaction torque for type k robots, x (i) s is the start state of robot i, and x (i) f is the desired state of robot i. In contrast to existing literature [6], we assume a tight spherical collision model and bound the interaction forces directly, eliminating the need of manually defining virtual collision shapes. For instance, larger f I(i) a,max and τ I(i) a,max will yield denser and more aggressive trajectories.…”

“…Often, a formation is restricted to 2-D planar motions [5]. For heterogeneous teams, even larger and asymmetric safety distances are required [6]. However, the downwash for two small Crazyflie quadrotors hovering 30 cm on top of each other is only −9 g, which is well within their thrust capabilities, and suggests that proper modeling of downwash and other interaction effects can lead to more precise motion planning and dense formation control.…”

Fast Generation of Collision-Free Trajectories for Robot Swarms Using GPU Acceleration

“…A further improvement in computational speed can by achieved by optimizing both trajectory and robot assignment in parallel, leveraging the fact that an optimal assignment of robots to goals will require far fewer collisions to be avoided, than a non-optimal assignment. This property is exploited by Agarwal and Akella [22], who reformulate the optimization problem to be solved as a linear sum assignment problem and apply their method to swarms of hundreds of robots; and by Preiss et al [1], Debord et al [23], and Hönig et al [24], who iterate a search-based roadmap planner with a trajectory smoothing step until feasible robot trajectories are found, and who demonstrate their method by planning collision-free trajectories for hundreds of robots through densely cluttered environments.…”