Session: MR-05/MSNDC-08-01 Motion Planning, Dynamics, and Control of Robots

Paper Number: 71602

Start Time: August 19, 10:00 AM

71602 - Robot Motion Planner for Under-Constrained Trajectories With Part-Specific Geometric Variances 

Industrial manipulators often interact with large and complex objects for a variety of automation tasks. Finding a feasible path for the robot end-effector that ensures task success is often non-trivial due to considerations such as reachability, singularity avoidance, and collision avoidance. This paper proposes an approach to expand the search space for feasible robot trajectories (and search for an optimal solution) by taking advantage of task redundancy for certain tasks while ensuring task completion. The effort builds on previous work enabling virtual fixture generation for complex shapes given CAD or scan data. The proposed method has been developed into a trajectory planning library on the ROS (Robot Operating System) framework and tested by simulating an interaction of a six-axis industrial robot with an aircraft fuselage. Results show increased coverage of task area with minimal robot base placements.

Automation of industrial and manufacturing processes with robots often involves tasks that necessitate planning around geometrically complex objects. Such tasks present a challenge of searching for feasible paths for the robot while avoiding singular configurations and collision with objects in its workspace. This problem is often exacerbated by over-constraining the trajectory requirements, since the exploration of a range of feasible paths or task completion strategies makes the problem too complex. In addition to the challenge presented by the search for a viable path, uncertainties in part placement often necessitate real-time adjustments to paths generated offline. The overall objective of this effort is to develop a trajectory planner with the ability to:

·        relax path requirements to facilitate the determination of valid trajectories that otherwise may not be possible to identify,

·        generate an optimal trajectory given the expanded trajectory search space,

·        make real-time modifications to pre-planned paths to address dynamic obstacles and part geometric variances.

The first two capabilities above are the focus of this paper.

Presenting Author: Ademola Oridate The University of Texas at Austin

Authors:

Ademola Oridate University of Texas at Austin
Mitchell Pryor University of Texas at Austin
Carolyn Conner Seepersad University of Texas at Austin