State estimation technique for a planetary robotic rover

Abstract

Given the long traverse times and severe environmental constraints on a planet like Mars, the only option feasible now is to observe and explore the planet through more sophisticated planetary rovers. To achieve increasingly ambitious mission objectives under such extreme conditions, the rovers must have autonomy. Increased autonomy, obviously, relies on the quality of estimates of rover's state i.e. its position and orientation relative to some starting frame of reference. This research presents a state estimation approach based on Extended Kalman Filter (EKF) to fuse distance from odometry and attitude from an Inertial Measurement Unit (IMU), thus mitigating the errors generated by the use of either system alone. To simulate a Sun-sensor based approach for absolute corrections, a magnetic compass was used to give absolute heading updates. The technique was implemented on MotherBot, a custom-designed skid steered rover. Experimental results validate the application of the presented estimation strategy. It showed an error range within 3% of the distance travelled as compared to about 8% error obtained from direct fusion.