Active sensing control improving SLAM accuracy for a vehicle robot.

This study considers an autonomous mobile vehicle using SLAM technology in an unknown environment. The focused problem is the decrease of trajectory following accuracy due to the decline of SLAM accuracy in an environment with insufficient information. To solve this problem, we propose to improve overall accuracy by controlling the vehicle's motion based on the SLAM accuracy. The motion control is implemented as a model predictive control that optimizes an evaluation function including the estimation accuracy. The estimation accuracy is evaluated as the Fisher information concerning the state, especially the velocity. The effectiveness of this method is verified by numerical simulation as compared with a trajectory tracking controller in the same environment. [ABSTRACT FROM AUTHOR]