The paper presents a new hybrid method, Neuro-Fuzzy Robot Route Planner D*Lite with Risk-Taking, for autonomous route planning of a mobile robot in an uncertain environment. This opens up new opportunities for the development of modern technologies in the operation of robotic systems. The proposed approach combines the D*Lite algorithm, neuro-fuzzy risk assessment, a LiDAR-based occupancy map, and a compact neural network TinyML-MLP for edge devices such as ESP32-Cam. The results of numerical simulations confirmed the ability of the method to form safe trajectories, maintain stable navigation performance, and ensure effective achievement of the target point in a dynamic environment.

Terms—autonomous navigation, mobile robot, modern technologies, neuro-fuzzy navigation, risk assessment, robotic systems, route planning