Nonholonomic motion planning in partially unknown environments using vector fields and optimal planners

Abstract

This paper presents a methodology to integrate vector field-based robot motion planning techniques with optimal trajectory planners. The main motivation for this integration is the solution of planning problems that are intuitively solved using vector fields, but are very difficult to be even posed as an optimal motion planning problem, mainly due to the lack of a clear cost function. Among such problems are the ones where a goal configuration is not defined, such as circulation of curves and road following. While several vector field based methodologies were proposed to solve these tasks, they do not explicitly consider the robot's differential constraints and are susceptible to failures in the presence of previously unmodeled obstacles. Our methodology uses a vector field as a high level specification of a task and an optimal motion planner (in our case RRT*) as a local planner that generates trajectories that follow the vector field and also consider the kinematic and dynamic constraints of the robot, as well as the new obstacles encountered in the workspace. To illustrate the approach, we show simulations with a Dubins like vehicle moving in partially unknown planar environments.