A VISION SYSTEM FOR A COGNITIVE ROBOT: COMBINATION OF MONOCULAR SLAM AND MEASURABLE MOVEMENT-CONTROL

摘要

One essential capability of an intelligent cognitive robot is detecting objects in a scene and capturing their details based on vision. In our project, the visual SLAM (simultaneous localization and mapping) with a single camera is applied to acquire sparse landmarks for a large scale map. As we adopt the monocular SLAM with a single camera, the measurable movement-control of the mobile robot is combined, which also enhances the accuracy and the robustness of the cognition process and helps with seizing the details of some small objects.