用于室外道路环境综合理解的多种新型视觉传感技术和系统

针对室外智能移动机器人自主导航的要求，提出了基于多种新型视觉传感技术的室外道路环境综合理解方法，其基本原理是综合利用机器人四周360°景物的环视图象信息、机器人前方道路的双目注视图象信息以及机器人运行过程中形成的时空全景图象信息，综合完成实时机器人行驶方向确定、实时路面障碍物检测和机器人全局定位等视觉任务。

Multiple Special-designed Visual Sensing Techniques and Systems for Comprehensive Understanding of Outdoor Road Environment

A new multiple resolution vision system is proposed for a mobile robot to navigate in an outdoor road environment. Three special-designed visual sensors meet the need for different visual tasks and are suitable to machine processing. The omni-directional imaging sensor acquires in real-time the images of the 360 degrees view around the robot and will not get lost when it acts as a spatial global visual sensor for road recognition. A single camera stereo sensor, which obtains in a snap the binocular images of the detail information in front of the robot, is dedicated to road obstacle detection. The spatio-temporal imaging method applied to roadside image sequences enables the robot to obtain temporal global information.and locates itself in the environment. Image-level invariance for each visual sensor are explored and the purposive visual tasks are fulfilled without any image segmentation and 3D recovery.