Bidirectional swarm search method for autonomous path planning

Abstract

Autonomous Path Planning (APP) problem is an important research topic in many fields including Mobile Robot (MR) applications. The main purpose of APP is to minimize the human intervention in searching feasible sequence path from the initial to goal position at optimal cost that satisfies any given constraints. Most of the paths planning algorithms developed so far used one direction information in developing the path solution. However, this approach leads to arguable solution. This research attempts to integrate the bidirectional searching strategy with Swarm Intelligence (SI) or called as Bidirectional Swarm (BiS) model. SI such as the Foraging Food Ant (FFAnt) behaviour is proven efficient in solving the path planning problem. The current research on FFAnt mostly focused on pheromone concept for agent communication. Instead, in this research the non-pheromone FFAnt is used and the agent communication is conducted via bidirectional interaction. The BiS was validated with double bridge experiment standard benchmark and similar result from the original double bridge model is obtained. The developed model has led to the development of Bidirectional Swarm based Path Planning (BiSPP) algorithm for MR using top-down methodology. The matrix performances of BiSPP are measured via computational time and path length. A series of experiment were conducted through the developed simulation tool with various static environments. The results are compared to Bidirectional Ant Colony Optimization algorithm and Multi-Scout Ants Cooperation algorithm. Results show that BiSPP algorithm has outperform the other two algorithms by decreasing up to 20 percent of the path length in a reasonable computational time. The simulation results indicate that the BiSPP algorithm has a potential to perform in static environment