Heuristic and Fitness-Based Path Optimisation for Maze Escape Robots

Abstract

Maze escape robots require efficient and reliable path planning to complete autonomous navigation in complex environments. This paper studies the path planning problem of maze escape robots, focusing on common limitations such as low search efficiency, redundant paths, long computation time, and poor adaptability in large or complex mazes. It first introduces the maze environment and robot model, where the maze is usually represented as a grid map with accessible areas and obstacles. Then, traditional algorithms, including DFS, BFS, and Dijkstra, are reviewed and compared in terms of search logic, path quality, and computational cost. Their limitations in dead-end avoidance, memory consumption, and large-scale search efficiency are discussed. On this basis, the paper analyses intelligent optimisation algorithms such as Genetic Algorithm and Particle Swarm Optimisation, which improve global search ability and reduce the risk of local optimum. An improved fusion strategy is further proposed by optimising the heuristic function and fitness model to balance path length, smoothness, and search efficiency. Simulation results show that the improved algorithm achieves shorter paths, faster calculation, and better stability in complex maze environments.