Selecting Sampling Based Planners for Local Planning in Unpredictable Environment for a Six-Axis Robot Manipulator

One of the solutions to planning in a changing environment involves two stages: global planning and local planning. This paper attempts to select a group of sampling-based planners for planning in a dynamic environment workspace for local planning according to the performance statistics of geometric sampling-based planners in a changing environment with unpredictable static objects. The planners are evaluated in a workspace with unpredictable static obstacles. The planners discern collision-free motion planning for a six-axis robot arm in changing environment. The robot arm has an eye-in-hand perception unit, the Realsense RGB-D camera, to model the surrounding. A static obstacle is introduced at random location at random time step in the workspace to emulate the changing environment in a work cell of an automation floor. The planners are subjected to an improved inverse-kinematic solver using a numerical method, the Newton-Raphson with sampling constraints, to facilitate online planning and replanning. The constraints are defined in the task space. The performance statistics are compiled for twenty four planners. In this paper, eleven planners are shown to reliably produce collision-free solutions at processing time of less than 0.1 seconds.