Open Access

Egehan Usluer

• This thesis explores how an already built Mini Mars Rover can autonomously navigate an uncharted maze, first by mapping it and then by applying different pathfinding strategies. The Mini Mars Rover uses IR sensors for wall detection, encoder-based odometry for tracking movement, an IMU for heading corrections, for input for more accurate localization. The strategy combines exploration techniques with practical path execution on real hardware, focusing on the challenges that come up when algorithms meet the physical world. Exploration applied by using a combination of Breadth-First Search (BFS) with the Travelling Salesman Problem (TSP) approach, as well as a Partially Observable Markov Decision Process (POMDP) strategy. Although both methods showed promising results in simulation, in hardware testing same results were not gathered. For the path planning, three pathfinding algorithms Flood Fill, Dijkstra’s, and A* are implemented and compared. The results of those algorithms and the strategy to adopt them into the robot is implemented. The project demonstrates a possibility of adaptation of those algorithms and their feasibility. The Mini MarsRover serves as a solid base for further development and improvements.