Algoritmos para camara de video

7th OAPS Working Paper Series Website: http://www.oaps.hk/

Paper No. 2010-007

An intelligent vehicle control system based on active vision
Xie Qihui1
1 2

Cheng Zhe2

School of Mechanical Engineering School of Naval Architecture, Ocean and Civil Engineering

ABSTRACT
In the Freescale Cup University Smart Car Racing Contest, many cars based on traditional passive vision lose theroad information because of the eye shot limitation of the camera. In order to solve this problem, this paper proposes an intelligent vehicle (IV) control system based on active vision by combining the angle of the camera with the posture of the front wheel. In this system, the IV turns the camera actively around the curve to find the guideline and make the road information be measured correctly. Thepaper gives an introduction to the innovative point of this proposal, i.e. the application of active vision in IV system, the significance of this design to the IV, and a series of problems and solutions during the process. The mechanical structure and control strategy of the final design are introduced as well. Experiments show that the active-vision system can evidently increase the stabilityand speed of the IV. Key words: Intelligent vehicle, Camera, Active vision, Image collection, Speed control

1 INTRODUCTION
Intelligent and information-based technology is the developmental tendency of the automotive industry, they have become a main indicator of the level of technological sophistication [1]. Practicality is the developmental tendency of intelligent vehicle (IV). An IV with goodadaptability will be the focus of research in the future. The best way to research IV based on vision navigation is experimenting on simulation models with similar mechanical architectures. So, with more and more attention being paid to research on IV and related fields, there are many smart model car contests aiming to improve college students’ ability of innovation and universities’ academiclevel in related field. The National University Smart Car Racing Contest hosted by Ministry of Education and co-sponsored by Freescale Semiconductor appears in this environment. With a background in rapid developing automobile electronics, the smart car racing contest covers technology innovation in subjects such as control theory, sensor technology, pattern recognition, electronic, electric,computer, mechanics etc. It aims to develop college students’ ability of innovation, study and research [2]. The rules of Freescale University Smart Car Racing Contest are as follows. Competitors
“上海大学生创新活动计划”项目（IAP2007）资助 1

Xie Qihui

Cheng Zhe

School of Mechanical Engineering

must use car models provided by the organizing committee, use Freescale Semiconductor’s 8-bit or 16-bitmicrocontroller as the core control module, design control strategy and system, including acquiring and processing of sensors’ signal, control algorithm, motor and drive module, steering control etc, independently. Competitors should make a model car that can recognize a path independently by adding road sensors, designing motor drive circuit, coding software and assembling model. After the system isdesigned and debugged, the car has to drive on a track, in middle of which lays a black guiding line. The winner is the one who finishes the race in the shortest time. The contest comprises of two groups, CCD and LED. We are part of the CCD group and use cameras to navigate. The control system of smart cars includes three parts: acquiring and processing sensors’ signal, speed control and steeringcontrol. The car alters speed and steering angle according to different road conditions after the road information is acquired by the camera and processed by the microcontroller [3-4]. Through statistical analysis of the tracks from all the contests, we can find that about 60%-70% of the track is curved. So, the speed and stability on the curves will largely determine the car’s performance. In the...