A Comparative Study of Pathfinding Algorithms for Low-Cost Mobile Robots in Dynamic Environments

Pathfinding is critical in mobile robotics for enabling autonomous navigation from a start to a goal location while avoiding obstacles. This study implements six representative pathfinding algorithms – Dijkstra, A*, Breadth-First Search (BFS), Greedy Best-First Search, Bug1, and Bug2 – and compares their performance on grid-based maps under low-cost robot constraints (limited battery and sensing) and dynamic changes (moving obstacles). We simulate a two-dimensional grid world with static and dynamic obstacles, modeling a simple wheeled robot with limited sensors and a finite battery. Each algorithm is evaluated on key metrics: path length, computation time, battery usage (proportional to distance traveled and actions taken), success rate (reaching the goal without failure), and adaptability to environmental changes. Our results show that A* consistently yields the shortest path and fastest search time in static, known environments, while BFS and Dijkstra also find optimal paths, albeit with higher computational costs. Greedy Best-First Search often finds a path quickly but can produce suboptimal or invalid paths under complex scenarios. The simple Bug algorithms (Bug1 and Bug2) are robust to unknown obstacles (requiring only local sensing) and guarantee finding a path if one exists, albeit at the expense of significantly longer detours and greater energy consumption. In dynamic scenarios (moving obstacles), global planners (A*, Dijkstra) must replan or may fail, whereas reactive Bug planners naturally cope by following obstacle boundaries. Overall, A* performs best in static settings with sufficient compute, while simpler methods or hybrid strategies may be preferable for very low-cost robots or highly dynamic settings. Our comprehensive comparison highlights the trade-offs of each algorithm and guides the choice of planning strategy based on environmental demands and resource constraints.