Application of Metaheuristic Search Algorithms for Path Planning of Smart Vehicles

Navigating a smart vehicle in an environment, determined or unknown, requires the localization of such vehicle in that environment using GPS, cameras, vision, laser or ultrasonic sensors, motion planning of the vehicle in free configuration space of the environment, and its ability to deviate from obstacles. In planning a path from a start to a goal configuration, the aim is to obtain the shortest path in less time while avoiding obstacles. Metaheuristic algorithms have been extensively applied to achieve this; while it exploits the environment from an initial solution, it also explores it to find a possible feasible path. Researchers in robotics and automation field have investigated and analysed the performance of population-based algorithms like Genetic Algorithm (GA), Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), Firefly Algorithm (FA), and Cuckoo Search Algorithm (CSA) to obtain a feasible shortest path. This research work investigates the performance of metaheuristic search algorithms like GA, PSO, FA, and CSA for path planning on four different benchmark problem maps (40 x 40m large, 20 x 20m maze, 20 x 20m rockpile and 20 x 20m pothole) and makes a comparative analysis based on computational time and path length. Furthermore, three sampling methods i.e., Random, Latin hypercube and pseudo-uniform sampling were used. It is observed that all the algorithms were able to achieve the optimal, although CSA performed poorly on path distance using GA and PSO on the maze and pothole map respectively, its performance on other two maps was relatively better both on shortest distance and computational time, however, it is concluded that no single algorithm is universally the best-performing algorithm for all maps. All simulations were performed on MATLAB R2022b. Keywords: Metaheuristic Algorithms, Path Planning, Navigation, Mobile Robot, Smart Vehicle