Because of light-time delays, a planetary rover located on Mars or beyond will probably need to be able to navigate autonomously, in order to do significant exploration. Recent work at JPL has explored several different autonomous navigation strategies. This work includes highly deliberative methods that require large amounts of computation and internal storage but yield very planful behavior to more reactive systems that require less resources but whose behavior is more difficult to model. This paper briefly presents these methods, the results from experiments both in simulation and in the field, and some conclusions on the value of different approaches, and their possible impact on the structure of the robot and the mission architecture.
[1]
Erann Gat,et al.
Exploiting known topologies to navigate with low-computation sensing
,
1991,
Other Conferences.
[2]
Larry Matthies.
Stereo vision for planetary rovers: stochastic modeling to near real-time implementation
,
1991,
Optics & Photonics.
[3]
Erann Gat,et al.
Path planning and execution monitoring for a planetary rover
,
1990,
Proceedings., IEEE International Conference on Robotics and Automation.
[4]
David P. Miller,et al.
Global Symbolic Maps from Local Navigation
,
1991,
AAAI.
[5]
David P. Miller.
Execution Monitoring for a Mobile Robot System
,
1990,
Other Conferences.