Autonomous navigation for Mars Rovers

Kurt Schwehr
NASA ARC Intelligent Mechanisms Group & CMU Robotics Institute

Friday, September 3, 1999
12:00 Noon—Pacific Forum

schwehr3.jpg (12292 bytes)The communications capability of planetary exploration vehicles is commonly constrained due to long transmission distances, limited communication windows, and restricted power usage. To effectively utilize robotic exploration, algorithms for autonomous navigation and data collection must be developed.

At Carnegie Mellon University, the Field Robotics Center has developed an integrated autonomous navigation system to support autonomous planetary robotic investigation techniques currently under development at other institutions.

The system integrates dead-reckoning, inertial navigation, and stereoscopic vision using the fast local obstacle-avoidance algorithm, Morphin, and the global dynamic-path- planning algorithm, DStar. The two algorithms work together using the Arbiter algorithm developed by the National Robotics Engineering Consortium.

The system was developed on an RWI ATRV-2 rover and will be tested this fall on the NASA/JPL 2005 Mars Rover prototype Fido/Rocky 8. The system is designed to run on a computer with the equivalent compute power and memory of a Pentium 133 with 32 Megabytes of RAM, a computer that is suitable for Martian deployment.

Next: Where the sea meets the sky- Recent studies of the wind-driven ocean

Last updated: December 19, 2000