An Improved Particle Filter Target Tracking Algorithm Using Online Learning Model

An improved particle filter target tracking algorithm is proposed which using optimization observation likelihood and Online Learning Model in this paper. Focusing on uncertain features, the detection of dynamic targets is realized by laser range finder observation, and an online learning algorithm of the proposed geometry model of the vehicle can automatically adapt to the dimensions of various vehicle body targets without any prior information. The experiment results on the Navlab and Standford Campus datasets show that the proposed method can achieve good autonomous navigation performance.