Changes in Path Tracking Performance of Autonomous Vehicle According to Design Feedback Gain and Look-ahead Distance of Pure-pursuit Algorithm

초록

Various researches on small autonomous driving platforms have been conducted for unmanned delivery or surveillance. Among the autonomous driving algorithms, the path tracking algorithm is the most basic algorithm for performing missions, but many vehicles have difficulty in ensuring tracking performance. The most commonly used pure-pursuit algorithm is difficult to tune and invests a lot of time because the two design parameters, such as feedback gain and look-ahead distance, have complex effects on performance. In this research, the tracking characteristics of pure-pursuit algorithm are identified, and the optimal controller conditions suitable for a small platform are determined. The tracking performance evaluation is conducted based on the ERP-42 autonomous driving platform. The controller design parameters are swept, and the maximum lateral and heading offset are measured under all parameter conditions.

키워드