Optimal robot base placement for robotic shipwelding

Sommer Gentry, Massachusetts Institute of Technology

Sandia National Laboratories’ AUTOGEN shipwelding project to date has solved some of the difficult problems of robot welding in generating automatic torch trajectories for tracing large weld lines along intersecting curved and flat surfaces given only CAD-specified geometries. However, the automation of torch welds will be incomplete until the problem of sequentially placing the robot base, or of moving it, so as to reach across large portions of the ship is solved. Indeed, even if only small sections of steel are welded at a time, optimal robot placement may improve the percentage of welds reachable by automation. The approach implemented in this work is based on Park and Cho [1], with the customization that instead of optimizing with the objective of connectably reaching the largest portion of a given trajectory with the robot in any permissible orientation, we optimize with the objective of covering the maximum expanse of weld in the configuration planned by the lower level AUTOGEN functions. The value of this approach, beyond integrating well with the current code, is that it demonstrates application of optimization even when we have no theoretical guarantees about the lower level planned robot joint movements.

1 Task oriented optimum positioning of a mobile manipulator base in a cluttered environment, Young Soo Park and Hyung Suck Cho. Journal of Intelligent and Robotic Systems, 18, n2, Feb 1997, pp 147-168.

Abstract Author(s): Sommer Gentry, Arlo Ames