The current paper focuses on the development of a Neural Network (NN) based hybrid adaptive controller。 A Thin Plate Spline (TPS) Radial Basis Function (RBF) is used as the NN activation function。 The TPS has distinct advantages over the other RBFs in terms of smoother interpolation with derivatives, better accuracy and faster convergence properties [8]。 Other RBFs cause combinatorial explosion, when the inputs are far away from the centered mean value, thus adversely affecting the controller performance。
The paper outlines the vehicle dynamic model [4, 7 and 9] first following which the development of control laws is detailed and finally the simulation results and conclusions are presented。 The current work is a continuation of the work already reported by the authors in [9]。 The NN based adaptive controller with the TPS RBF has better cruising and lane following performance characteristics compared to a conventional PD controller。
All rights reserved。 No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of TTP, www。ttp。net。 (ID: 203。199。213。66-15/07/14,05:58:20)
Vehicle Dynamics
Various model parameters used in the vehicle dynamic model are adapted from the Tata Indica passenger car platform (Tata Motors limited, India) [10]。 The different forces acting on the car are shown in Fig。 1 and Fig。 2。 The vehicle dynamics model is developed based on the assumption that the roll, pitch and bounce motions are negligible。 The present model does not take into consideration the effect of the vehicle suspensions on the wheel axles。
Fig1。 Free body diagram of the vehicle
Fig。 2。 Vehicle dynamics model
The equations for the vehicle dynamics model are derived using the Newton-Euler formulation [7, 9]。 The balance of forces and moments along the longitudinal axis and lateral axes are given by Eq。1- Eq。3。
) (mg - k x˙ 2 )) sinδ - C (
- δ) cos δ - m x˙ θ˙ - C ( y˙ − bθ˙ ) 。 (2)
The equations Eq。1 – Eq。3 can further be reduced to a matrix form, Eq。4, when steering angle δ is assumed to be very small (small angle approximation) i。e。 the values of cos δ and sin δ in the above equations can be approximated to 1 and δ respectively。
v (q˙ , u) = [vx vy vz ] and u = [F δ]
The control input vector u is determined from the controlled terms v (q˙ , u) , an output from the controller。
Development of Hybrid Adaptive Combined Controller
The autonomous driving controller has two main objectives, namely: maintaining the inter-vehicle spacing and control of lateral deviation from the road center line。 The lateral and longitudinal errors in terms of: leader vehicle’s position (xLeader), desired inter-vehicle spacing (xspacing), desired lateral deviation (yd) and position of the follower vehicle (x, y) are given by Eq。5 and Eq。6。摘要：本文介绍了自主循迹智能车辆的设计。软硬件的设计方法是利用反射型红外光电传感器作为路径识别模块来实现自主循迹功能。该系统采用飞思卡尔HCS系列16位单片机MC9SDG128作为其主控制器,利用光电传感器阵列来识别路径信息。单片机激励伺服电动机运转，并根据从传感器中获得的路径和速度信息分析控制直流电动机的速度。因此，光电传感器自主循迹原理的提出，使这种智能车辆可以跟踪黑色导航线自动地快速顺畅的沿线行驶。并且还对传感器的空间分布对自主循迹的影响进行了讨论。
关键词：光电传感器；智能车辆；自主循迹；路径识别和控制
一、引言
智能车，又叫轮式移动机器人，是一个综合性的具有环境感知，路径规划和自动交通功能的智能系统。作为一个高科技产品，智能车辆集合了自动控制，模式识别，传感技术，汽车电子，电气，计算机，机械等诸多学科技术。随着自动化，计算机和信息技术的发展，智能车在工业和日常生活中扮演着非常重要的角色。例如，用于太空探索的月球探测车，在图书馆或车间的搬运机器人等。近年来，智能小车被广泛应用于道路，物流和灵活的制造业系统，并且成为了人工智能领域研究的热点。 红外光电传感器的智能循迹车辆英文文献和中文翻译(7):http://www.youerw.com/fanyi/lunwen_100371.html