INTELLIGENT PATH SEARCHING AND POSE ESTIMATION

Autonomous vehicles have been witnessed prosperous in the last decade. This thesis focuses on path searching and pose estimation for autonomous vehicle navigation. The thesis first studies multi-sensor pose estimation on the vehicle that is equipped with cameras, an altitude and heading reference system (AHRS), and digital maps. The proposed pose estimation has been validated using public and self-collected data and can be further extended to other kinds of sensor configurations with state and measurement constraints. Then, maximum entropy searching and interactive obstacle avoidance have been proposed for global and local path searching, respectively. The maximum entropy searching has been introduced as a new heuristic of exploration especially in complex and unstructured environments to improve naive goal-reaching and random searching techniques. Based on personal space assumption, artificial potential fields have been redesigned such that human’s emotion detected from facial expressions is taken into consideration in obstacle avoidance.

Author dc:creator

• JIANG RUI

• path searching, pose estimation, autonomous vehicles