摘要
提出一种双足步行机器人的实时障碍检测视觉系统.基于图像平面与机器人行走地面之间的映射变换矩阵的唯一性准则,判别图像中的像点是否位于地面上,高于或低于地面的点被认为是障碍点.为减少实际行走过程中系统外部和内部参数变化对映射变换矩阵的影响,系统加入了在线校正映射变换矩阵模块.在提取出障碍物体边缘后,通过对其三维信息的简单恢复,建立了机器人行走空间的障碍投影图.该系统计算量小,可靠性强,能基本满足双足步行机器人实时避障的要求.
A real time obstacle detection system that uses a stereo pair of cameras but needs no 3D reconstruction of the surroundings is presented. At first, the projective transformation matrix between the image plane and the ground plane on which the biped robot walks is established. Next, planar projection stereopsis method is used to decide whether or not each point in a stereo image exists on a ground plane. In order to decrease the effects on the projective transformation matrix caused by the variety of outer or inter system parameters, an on-line adjusting module is adopted. After extraction the edge map of obstacles, a free-space map can be created by simple recovery to 3D information of obstacles. The system is simple and robust enough to make the robot avoid obstacles in time.
出处
《控制与决策》
EI
CSCD
北大核心
2004年第1期40-43,48,共5页
Control and Decision
