Mesh Manifold Based Riemannian Motion Planning for Omnidirectional Micro Aerial Vehicles

This letter presents a novel on-line path planning method that enables aerial robots to interact with surfaces. We present solution the problem of finding trajectories drive robot towards surface and move along it. Triangular meshes are used as map representation is free fixed discretization allows for very large workspaces. propose leverage planar parametrization methods obtain lower-dimensional topologically equivalent original surface. Furthermore, we interpret its manifold approximations allow use Riemannian Motion Policies (RMPs), resulting in an efficient, versatile, elegant motion generation framework. compare against several Rapidly-exploring Random Tree (RRT) planners, customized CHOMP variant, discrete geodesic algorithm. Using extensive simulations on real-world data show proposed planner can reliably plan high-quality near-optimal at minimal computational cost. The accompanying multimedia attachment demonstrates feasibility real OMAV. obtained paths less than 10% deviation from theoretical optimum while facilitating reactive re-planning kHz refresh rates, enabling flying perform interaction complex