Design and Implementation of a Novel Weighted Shortest Path Algorithm for Maze Solving Robots

This research presents design and implementation of the shortest path algorithm for labyrinth discovery application in a multi-agent environment. Robot agents are unaware of the maze at the beginning, they learn as they discover it. Each agent solves a part of the maze and updates the shared memory so that other robots also benefit from each other's' discovery. Finding of the destination cell by an agent helps others to interconnect their discovered paths to the one ending with the destination cell. The proposed shortest path algorithm considers the cost for not only coordinate distance but also number of turns and moves required to traverse the path. The Shortest Path algorithm is compared against various available maze solving algorithms including Flood-Fill, Modified Flood-Fill and ALCKEF. The presented algorithm can be used also as an additional layer to enhance the available methods at second and subsequent runs.