Sequential Bayesian Estimation of Course Block Position and Velocity of mini 4WD AI Car

Abstract

Competitions in Mini 4WD AI tournament are held from three points of view: (1) high-speed time attack, (2) accuracy of stopping position, and (3) running close to the specified time, In (1), vehicle’s position and speed are estimated, and speed control is performed to prevent deviation from the target while driving at high speed. In (2), apply brakes and stop just before the finish line while counting the number of laps. In (3), strategy is to run at a constant speed and adjust the speed according to the remaining time. Where, in a vehicle traveling at high speed, the communication rate is slow compared to the required sampling interval, and it is necessary to estimate the vehicle’s own state from insufficient sensor information. In this paper, The self-position is estimated by simplifying it to a discrete position of the course block instead of a continuous value. Furthermore, the effectiveness is evaluated through state estimation simulations including speed.