Intelligent Parallel Parking Using Adaptive Neuro-Fuzzy Inference System Based on Fuzzy C-Means Clustering Algorithm

Today’s intelligent self-driving vehicles alongside technology development are more believable. One of the intelligent features of self-driving cars is autonomous parking which has been specifically considered in industry and academic research areas. This paper focuses on the autonomous parallel parking. First, the vehicle kinematics modeling by considering Ackermann angle calculation has been thoroughly explained and then the desired path by satisfying spatial conditions and its proportional steering angle is extracted. Autonomous parking scenario has been defined based on two phases of forward and backward motions. Accordingly, the desired training data includes steering angle and vehicle motion feedbacks (x, y, φ) that are utilized for designing intelligent controller. The proposed control system has two levels: upper and lower level. The former is a supervisory controller which switches between phases while the latter controls the vehicle based on received feedbacks from sensors in each phase. In this research adaptive-network-based fuzzy inference system (ANFIS) based on fuzzy c-means clustering (FCM) is employed to model the expert driver as an intelligent controller in parking maneuver. In this structure, FCM is used to systematically create the fuzzy membership functions and rule base for ANFIS. The performance of the proposed control algorithm is verified by defining an accuracy index. The simulation results in three different constant speeds indicate the value of accuracy index and jerk of controller output signals remains in an acceptable band.