UAV Path Planning Using Informed Search Techniques on Custom Flight Controller

Autonomous drones are pivotal in various applications like surveillance, delivery of goods, emergency rescue etc., emphasizing the need for seamless navigation. This research centers on running path planning algorithms on UAVs to facilitate autonomous and optimal route determination. Evaluating A-star(A*), and Greedy Best-First Search (GBFS), we highlight A* ‘s superiority in computational efficiency and path optimality. To bridge the gap between real-world terrains and algorithmic requirements, we propose a novel integration of computer vision, transforming geographical maps into adaptive grids for UAVs. Path planning is done on a pre-shot image and the optimal navigation path for the UAV is calculated. In a real-time system, a camera is used to capture video continuously of the environment and the flight controller works on feature extraction and classifies obstacles frame by frame (typically using a software like Open-CV). In our case, the implementation was done on a custom flight controller based on the ESP platform which uses a Tensilica processor. Our findings demonstrate the efficiency of this approach, and offer a promising solution for autonomous UAV navigation in dynamic environments.