Pheromone Accumulation and Iteration

Abstract

In this chapter, the robot path-planning problem is explored under the vision–brain hypothesis, and meanwhile, the pheromone accumulation and iteration mechanisms and processes are explicitly illustrated. Based on the hypothesis, robots can recognize obstacles, and therefore, to solve the robot path-planning problem, it remains to decide the optimal path to the targets or the regions of interest. Differing from most studies on the robot path planning, the significance of pheromone paths (sub-paths) in full path generation is emphasized, employing the ant colony algorithm, where pheromone updates are directed through calculating the passed ants of the sub-paths in each iteration. This algorithm can be further improved by placing pheromone on the nodes to improve the efficiency of the pheromone storage and updates, where the ant colony (a series of pheromone points) becomes a pheromone trace. Utilizing localization rules and one-step optimization rules for local optimization, the time to construct the first complete solution can be shorten and a better solution of the problem of the robot path planning can be generated by establishing a mesh model of the navigation area with determined obstacles. Utilizing the locally compressive sensing algorithm in Chap. 2 and selecting a behavior-sensitive area for compressive tracking, machine can recognize some special global behaviors (e.g., running, falling) and local behaviors (e.g., smiles and blinking) and the recognition rate and accuracy can be ensured. The broad learning system with a vision–brain (the decision layer) introduced in Chap. 2 for face recognition is further utilized to tackling a series of challenging issues—illumination changes, expression and pose variations and occlusion problems, respectively, utilizing some representative face databases. Results show that face recognition rates in 100 times of training and testing on each database can approach to 100%, including the database with real disguise occlusion.