One of the biggest challenges was replacing our previous left-wall-following approach with a more intelligent navigation system. During nationals, the robot became trapped in loops on floating tile layouts, making it clear that a reactive algorithm alone wasn't sufficient. To solve this, I developed and integrated a hybrid Depth-First Search (DFS) and Breadth-First Search (BFS) navigation system, allowing the robot to explore unknown mazes efficiently while always maintaining a valid path back to the starting tile.

Another major challenge was integrating multiple subsystems into a single reliable control loop. Navigation, obstacle detection, ramp and staircase traversal, victim detection, tile identification, and rescue kit deployment all needed to operate simultaneously without interfering with one another. This required extensive debugging of sensor timing, state management, and communication between modules to ensure the robot could make consistent real-time decisions.

Finally, competition environments are unpredictable. Sensor readings varied under different lighting conditions, wheel drift accumulated over long runs, and physical obstacles introduced alignment errors. Through iterative testing and calibration, we refined the robot's navigation, improved sensor reliability, and built recovery behaviours that allowed the robot to adapt rather than fail when conditions changed.