Multi-robot Visual Tracking System based on Cognitive Sharing of Invariants

点击翻译按钮获取中文摘要

In multi-robot systems, since each robot has generally a different perspective around an object to be shared, an effective representation for cognitive sharing would be different. This work proposed a hierarchical cognitive invariant model as a framework of cognitive sharing for multi-robot system. The ambiguity measure of the effective representation of the shape or color feature is dynamically evaluated according to the change of perspective of each robot. Relational representation of the object is formed by autonomous landmark generation called representation triangle, which specifies the existence region of target object. A visual mobile object tracking with multi-robot was experimentally demonstrated to show the effectiveness of proposed method where an occlusion occurs and a lightning condition changes.