The navigation system of the JPL robot

  • Authors:
  • Alan M. Thompson

  • Affiliations:
  • Jet Propulsion Laboratory, Pasadena, California

  • Venue:
  • IJCAI'77 Proceedings of the 5th international joint conference on Artificial intelligence - Volume 2
  • Year:
  • 1977

Quantified Score

Hi-index 0.02

Visualization

Abstract

The control structure of the JPL research robot and the operations of the navigation subsystem are discussed. The robot functions as a network of interacting concurrent processes distributed among several computers and coordinated by a central executive. The results of scene analysis are used to create a segmented terrain model in which surface regions are classified by traversibility. The model is used by a path-planning algorithm, PATH, which uses tree search methods to find the optimal path to a goal. In PATH, the search space is defined dynamically as a consequence of node testing. Maze-solving and the use of an associative data base for context-dependent node generation are also discussed. Execution of a planned path is accomplished by a feedback guidance process with automatic error recovery.