Hybrid Path Planner for Centralized Multi-Robotic Long-vehicle based on Adaptive Dimensionality and Grey Wolf Algorithm

The present robot planning methods pay no attention to the impact of the robot's size and the space it occupies in the environment on path planning. This paper presents a centralized hybrid method to plan optimal paths for multiple robotic long vehicles (RLVs) competing with each other to reach one common goal, taking into account the space that must be available to the RLV at each step to avoid narrow spaces that are too small to pass through. The environment analysis for each RLV is improved by assigning constant weight to the obstacles and calculating two new parameters: safety (SF) and objective (OBJ) to give an initial evaluation for each position (node). Safety is the summation of obstacle weights (W), and objective is the summation of cost and safety. The analyzer also detects narrow free spaces that are smaller than the vehicle size and changes their type to (trap) to avoid them later during movement. In the movement stage, adaptive dimensionality (AD) finds valid nodes that meet conditions of range of steering angle and connectivity. Among the valid nodes, grey wolf optimization (GWO) finds the leaders, i.e., path nodes discovered within the sensing window. The proposed method is simulated by computer in different dynamic environments, and the results have proved that the method succeeds in leading the robots to the goal with the safest path. It was observed that increasing the number of RLVs leads to an increase in planning time.