The invention discloses a robot vision positioning system. The system comprises the following components: a video acquisition module, a video angle adjustment module, a data processing module, a three-dimensional reconstruction module, a coordinate calibrating module, a route outputting module, and a robot driving module. A kinect depth sensor is used for acquiring a target position depth image and a barrier depth image, and reconstructing a three dimensional image, a target position in the three dimensional image, coordinate positioning of a barrier and real time positioning data of the robotare used for outputting a target route, in order to realize real time update of the target route, and improve accuracy of positioning results; an experiment shows that different depth images acquiredby the Kinect sensor are used for completing three dimensional reconstruction of the target object, only 48 seconds are needed, reconstruction effects are very fine, and efficiency of vision positioning is substantially improved.