基于粒子滤波的仿人机器人跟踪算法研究

Study of Humanoid Robot Tracking Based on Particle Filter

针对RoboCup标准组仿人足球机器人NAO的特点,在研究传统粒子滤波跟踪算法基础上,提出基于颜色特征和改进的粒子滤波相结合的机器人跟踪算法。改进的粒子滤波算法在传统采样重要性重采样(SIR)算法重采样过程中有选择地增加一组随机粒子,解决了传统采样重要性重采样(SIR)算法因被跟踪运动目标运动不规则而出现跟丢的问题。实验结果表明,该算法增强了基于SIR的粒子滤波跟踪算法的适应性和有效性。

Aiming at the special problems of the humanoid robot NAO in RoboCup environment, based on the research of general Particle Filter（PF） tracking algorithm, a new robot tracking algorithm combined improved PF with color character is presented. The improved PF tracking algorithm selectively increases a group of stochastic particles during the Sampling Importance Resampling（SIR） process based on traditional PF algorithm, and the improved algorithm can track the robot which movement is in-cgular. Finally, the experiments show the presented algorithm is efficient and robust.