Study on Geometric Path Tracking Algorithm for Tracked Vehicle Model

This paper proposes a geometric path tracking algorithm for mining robot. In Korea, the pilot mining robot has been developed for collecting manganese nodules from deep seabed. The pilot mining robot demands a localization algorithm to determine an accurate position and an automatic path tracking algorithm. The fundamental performances of two algorithms are verified by using a ground tracked vehicle to reduce the time and cost for in-situ test. Ultrasonic position system (UPS) is selected to measure the position. UPS is very similar with the measuring principle of LBL. The localization is realized by data-fusion of UPS (data period of 0.6sec) and motor encoder (data period of 0.1sec). It focuses on how to decide weighting values of two position data having different periods in order to produce better position information. For tracking the specified path, two algorithms, the follow-the-carrot and the pure pursuit, are comparatively investigated. The follow-the-carrot algorithm finds a line passing between the center point of vehicle and the target point, and controls the vehicle to follow the line with constant speed. The pure pursuit algorithm is that vehicle is controlled through radius of a circle contacting with forward vector of the vehicle and passing between the center point of vehicle and the target point. The appropriate algorithm for driving on soft-soil can be selected and proven by numerical simulation and experiment.