Adaptive neuro-fuzzy control of autonomous ground vehicle (AGV)‎ based on machine vision

Abstract EN

Development of self-autonomous robot that has the capability to deal with dynamic obstacles as well as static obstacles and keep finding its optimal path represent big challenge to many of the researchers nowadays.

Utilization of artificial intelligence (AI) in developing a robust algorithm for nonlinear system was proved in some applications.

The present work focus on excessive analysis and investigation of the optimal AI algorithm technique for controlling the path planning of four -wheel autonomous robot to achieve its set target through both static and dynamic obstacles with the optimal path without human interference.

Fuzzy logic, Neural Network, and Adaptive Neuro-Fuzzy Inference System (ANFIS) were tested and evaluated as controlling algorithm for this navigation task.

The performance evaluation was done both theoretically and experimentally using vision-based feedback sensor, Kinect camera, on a real four-wheel mobile robot.

Results prove that ANFIS successfully control the automatic path planning of the robot with the optimal path by reaching its target with % 48.01 less than Neural Network and % 58.51 less than Fuzzy Logic.

ANFIS controller also has given safer path as well as flexible obstacles avoidance through its path to final destination with smooth time-efficient path in the cluttered environment.