Path planning approach for ground robots in 3D environments using sampling-based algorithms

Abstract

Path planning is one of the fundamental problems in robotics. Due to advancements in technology, the application of mobile robots has increased in recent years. This is, not only in the field of robotics but in other domains such as computational biology, computer animation, and aerospace. Path planning in high dimensional environments for mobile robots is known to be computationally challenging, but since the introduction of the sampling-based planning algorithms such as rapidly exploring random tree (RRT) and probabilistic roadmap (PRM), solving high dimensional path planning problems has become easier. In this dissertation, a mesh-based RRT path planning approach is presented. Connecting RRT tree nodes on a nonplanar surface mesh requires the computation of geodesics, shortest length paths on the mesh, which can create a high computational load. The proposed method reduces the number and length of geodesics on the mesh. Simulation results show that this method finds a feasible path faster than the basic RRT on a synthetic mesh surface. Operation is also shown on a real mesh surface.