Abstract

This paper presents the development and simulation results of a Heuristic Application-Specific Path Planner (HASPP) that can be used to automatically plan trajectories for a manipulator operating around obstacles. Since the implementation of HASPP is inherently application-specific due to dependence on heuristics, the application of HASPP to an eight degree of freedom Pipe Manipulator is presented as an illustrative example. This development and simulation was implemented on a Silicon Graphics Personal IRIS with the aid of WALKTHRU, a 3-D simulation and animation tool, and software developed in C. HASPP uses extensive knowledge of the manipulator’s workspace and makes certain assumptions about the environment in finding trajectories. The algorithm also makes use of the manipulator’s redundant degrees of freedom to avoid obstacles and joint limits during the trajectory while obtaining a heuristic near-optimal solution. The algorithm is rule-based, governed by heuristics and well-defined geometric tests, providing extremely fast results. It finds “good” trajectories that are optimal within the defined heuristics. When a trajectory is not feasible for the given geometry, the algorithm offers a diagnosis of the limiting constraints.

The Pipe Manipulator HASPP implementation has been tested thoroughly with the computer graphics model and it has demonstrated the ability to reliably determine near-optimal collision-free erection trajectories completely automatically. No other planning techniques available in the literature have demonstrated the ability to solve problems as complex as the example presented here. The use of HASPP with simulation offers many application opportunities including plant design constructability studies, assembly and maintenance planning, pre-planning and pre-programming of equipment tasks, and equipment operator assistance. This work was the result of construction automation research sponsored by the National Science Foundation.

This content is only available via PDF.
You do not currently have access to this content.