以DSP平台實現RTK-DGPS為基礎之車輛自動駕駛系統

Abstract

智慧型運輸系統(ITS)已成為世界先進國家的重點發展項目。自動駕駛(Automatic Driving)已經是先進車輛控制及安全系統(AVCSS)相關領域主要的趨勢。本研究使用數位訊號處理器C6713 DSK為平台，實現以即時動態差分全球衛星定位系統(Real-Time Kinematic Differential Global Positioning System，RTK-DGPS)為主要的導航定位輸入並使用雷射測距儀(Laser Range Finder)來達到防碰撞功能的嵌入式自動駕駛系統。RTK-DGPS設備可以達到大約2公分的定位精度，足以使車輛自動導航有很好的控制。在橫向控制方面使用模糊邏輯(Fuzzy Logic)計算方向盤所需的轉角，並使用PD控制方法使致動器達到所需的位置。縱向控制方面根據道路狀況指定所需的速度，並使用速度管理器(Supervisory Controller)優化目標速度，在致動器控制方面使用調節控制器(Regulation Controller)管理油門煞車的致動。在安全方面，於車頭安裝雷射測距儀偵測前方障礙物，透過速度管理器(Supervisory Controller)所計算出來的安全速度達到安全的自動駕駛。系統安裝在實驗車TAIWAN iTS-1上作測試，實驗的結果展現很好的自動駕駛性能，而且與人為開車的駕駛習慣相似。

Intelligent Transportation Systems (ITS) have become a key constructive item in many developed countries. Automatic driving has been a main trend in the field of advanced vehicle control and safety system (AVCSS). The C6713 DSK is served as a platform in the present study to implement an embedded system for automatic driving by using a real-time kinematic differential global positioning system (RTK-DGPS) as the main sensor input for positioning and using a laser range finder as the sensor for collision avoidance. The RTK-DGPS equipment can reach a 2-centermeter positioning accuracy which can finely locate the vehicle trajectory. In lateral control, fuzzy logic is used to calculate the target position of the steering wheel and PD control is used to move the steering bar by means of an actuator to achieve the target position. The vehicle speed in longitudinal control is according to the road condition and supervisory controller is used to optimize the target speed. As for the actuator control, regular controller is adopted to manage the actuation of brakes. For the safety, the head of the car is equipped with a laser range finder to detect obstacles in the front, though the safe speed calculated by supervisory controller to make a safe automatic driving. It is tested in the experiment car, TAIWAN iTS-1. The experimental results show that the combination of RTK-DGPS and artificial-intelligence-based techniques has a good performance in automatic driving, and the resulted steering maneuvering is very similar to human driving.