Parking motion planning and control for motor vehicles using a neural network with additional rules

Motoji Yamamoto, Katsuki Ohba, Akira Mohri

Research output: Contribution to journalArticle

Abstract

This paper addresses the issue of parking motion planning and control of motor vehicles with front steering. Because of the non holonomic constraints on kinematics of the vehicles, parking motion control is a difficult problem. A neural network control system is proposed for such problems. The system learns parking control strategy using good human driving data of parking. The perfect learning of neural networks using actual driving data is not so easy. Therefore, some additional rules are also proposed to refine the parking motion for the control system by the neural network. The resulting control system consists of a neural network and additional rules. The system is tested in a restricted area surrounded by walls as a motion planner and a motion controller. The simulation results show that the system works well for the learned wall shapes and for another wall shapes slightly different from learned ones. An experimental result also shows the effectiveness of the system.

Original languageEnglish
Pages (from-to)2172-2177
Number of pages6
JournalNihon Kikai Gakkai Ronbunshu, C Hen/Transactions of the Japan Society of Mechanical Engineers, Part C
Volume62
Issue number598
Publication statusPublished - Jun 1996

    Fingerprint

All Science Journal Classification (ASJC) codes

  • Mechanical Engineering

Cite this