300 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 5, NO. 4, DECEMBER 2004 Springrobot: A Prototype Autonomous Vehicle and Its Algorithms for Lane Detection Qing Li, Nanning Zheng, Senior Member, IEEE, and Hong Cheng Abstract—This paper presents the current status of the Springrobot autonomous vehicle project, whose main objec- tive is to develop a safety-warning and driver-assistance system and an automatic pilot for rural and urban traffic environments. This system uses a high precise digital map and a combination of various sensors. The architecture and strategy for the system are briefly described and the details of lane-marking detection algorithms are presented. The R and G channels of the color image are used to form graylevel images. The size of the resulting gray image is reduced and the Sobel operator with a very low threshold is used to get a grayscale edge image. In the adaptive randomized Hough transform, pixels of the gray-edge image are sampled randomly according to their weights corresponding to their gradient magnitudes. The three-dimensional (3-D) para- metric space of the curve is reduced to the two-dimensional (2-D) and the one-dimensional (1-D) space. The paired parameters in two dimensions are estimated by gradient directions and the last parameter in one dimension is used to verify the estimated pa- rameters by histogram. The parameters are determined coarsely and quantization accuracy is increased relatively by a multireso- lution strategy. Experimental results in different road scene and a comparison with other methods have proven the validity of the proposed method. Index Terms—Autonomous vehicle, lane-boundary detection, machine learning, randomized Hough transform (HT). I. INTRODUCTION A UTOMATIC vehicle driving aims at automating (entirely or in part) various driving tasks. The tasks of automati- cally driven vehicles include road following and keeping within the correct lane, maintaining a safe distance between vehicles, regulating the speed of a vehicle according to traffic conditions and road characteristics, moving across lanes in order to over- take vehicles and avoid obstacles, searching for the correct and shortest route to a destination, and driving and parking within urban environments [1]. The field of intelligent vehicles is growing rapidly in the world, in terms of both the diversity of emerging applications and the levels of interest among traditional players in the au- tomotive, truck, public transportation, industrial, and military communities. Intelligent vehicle (IV) systems offer the potential for significant enhancements in safety and operational effi- ciency. As one component of intelligent transportation systems Manuscript received November 30, 2003; revised July 15, 2004 and August 3, 2004. This work was supported by the National Natural Science Foundation of China under Grant 60205001 and Grant 60021302 and the “211” Key Subject Foundation, Xi’an Jiaotong University. The Associate Editor for this paper was S. Tang. The authors are with the Institute of Artificial Intelligence and Robotics, Xi’an Jiaotong University, Xi’an 710049, China (e-mail: qli@aiar.xjtu.edu.cn; nnzheng@mail.xjtu.edu.cn; hcheng@aiar.xjtu.edu.cn). Digital Object Identifier 10.1109/TITS.2004.838220 (ITS), IV systems understand the surrounding environment by a different sensor combined with an intelligent algorithm in order to either assist the driver in vehicle operations (driver assistance) or fully control the vehicle (automation) [2]. For solving the problem of traffic congestion and acci- dents, especially in big cities such as Beijing, Shanghai, and Guangzhou, the Chinese government has been increasing funding for improving the traffic infrastructure, enforcing traffic laws, and educating drivers about traffic regulation. In addition, research institutes have launched research and development projects in driver assistance and safety warning systems [3]. The Institute of Artificial Intelligence and Robotics, Xi’an Jiaotong University, Xi’an, China, is developing an autonomous vehicle called Springrobot. The research mainly covers the basic vehicular dynamic behaviors, the vehicular sensory technology and in-vehicle data communication, signal processing and the design of information fusion for monitoring running conditions and vehicle–road interactions, the driver-assistance and safety- warning systems, and full control of the vehicle and mobile- agent technology. As one of the intelligent functionalities, lane detection is the problem of locating lane boundaries. Up to until the present, various vision-based lane-detection algorithms have been de- veloped. They usually utilized different lane patterns (solid or dashed white painted line, etc.) or different road models [two-di- mensional (2-D) or three-dimensional (3-D), straight or curved], and different techniques (Hough, template matching, neural net- works, etc.). There has been a significant development of research on ma- chine vision for road vehicles [4] and the lane-boundary detec- tion has been an active research area over the last decade. An automatic lane detector should be able to handle both straight and curved lane boundaries, the full range of the lane mark- ings (either single or double and solid or broken), and pavement edges under a variety of types, lane structures, weather condi- tions, shadows, puddle, stain, and highlight. Many systems of lane-boundary detection have been reported in [5] and [6]. In [7], the authors proposed a B-Snake-based lane-detection and tracking algorithm without any cameras’ pa- rameters. The B-Snake-based lane model is able to describe a wider range of lane structures, since B-Spline can form any ar- bitrary shape by a set of control points. The problems of de- tecting both sides of lane markings (or boundaries) have been merged as the problem of detecting the midline of the lane, by using the knowledge of the perspective parallel lines. The Canny/Hough Estimation of Vanishing Points (CHEVP) is pre- sented for providing a good initial position for the B-Snake and minimum mean-square error (mmse) is proposed to determine 1524-9050/04$20.00 © 2004 IEEE