Real‐time multiple mobile robots visual detection system

The purpose of this paper is to describe how the authors designed a small satellite formation ground test bed in order to study the small satellite formation flying technologies, such as autonomous formation control and network communication. As one of the subsystems, the vision detection system is responsible for the pose (position and orientation) detection of the three small satellite simulators, each of which is composed of a wheeled mobile robot and an on‐board micro control unit. In this paper, the rapid vision locating of the three small satellite simulators in the wide field is discussed.

The scene size required by the test bed has exceeded the scope of one camera, thus how to obtain the complete scene becomes a difficulty. On the base of image mosaic, a vision system composed of two cameras is designed to capture the scene simultaneously. After the two overlapped images are rapidly stitched, the real‐time view of the big scene is attained. Second, the new color tag representing the pose of small satellite simulators is designed, which can be easily identified.

A real‐time multiple mobile robots visual locating system is introduced, in which the global search algorithm and track search algorithm are combined together to identify the real‐time pose of multiple mobile robots. The switching strategy between the two algorithms is given to ensure the accuracy and improve retrieval speed.

The paper shows how, without camera calibration, the pose of each small satellite simulator in the world coordinate system can be directly calculated by the coordinate transformation from the image coordinate system to the world coordinate system based on relative measurement. The accuracy and real‐time performance of the vision detection system have been validated by experiments on locating static tags and dynamic tracking three small satellite simulators.