Multi-sensor Particle Filter Based on Improved Grey Wolf Optimizer

The particle degeneration is a significant factor effecting filter performance in traditional particle filter, which poses a challenge for its application in multi-sensor fusion. The resampling is a valid scheme to counter this issue, such as roughening, a prior editing procedure, and regularisation, but these measures are prone to generate the sample impoverishment. Different from these methods, this paper proposes a multi-sensor particle filter based on improved grey wolf optimizer (IGWO) from the perspective of intelligent optimal sampling. Firstly, the grey wolf optimizer is modified to acquire better exploitation and exploration abilities. Then, the IGWO is used to optimize the posteriori distribution of particles through guiding them to move towards the high likelihood region, which not only avoids the particle degeneration, but also holds the diversity of particles. Finally, a numerical example that an indoor localization for mobile robot with multi-camera visual system is provided to illustrate the effectiveness of the proposed algorithm.