A Vision-Based Perception Method for External State of Spatial Cellular Robot

Abstract

The invention provides a vision-based method for sensing the external state of a space cell robot, which belongs to the field of space cell robot distance measurement. The present invention establishes the image database of different types of cells of the space cell robot, performs image labeling on the connection surface that needs to be identified and autonomously connected to the database image, and performs network training based on deep learning on the marked image, and uses the connection surface in the pixel coordinate system The position in the camera, as well as the conversion relationship between the pixel coordinate system, the image plane coordinate system and the world coordinate system, combined with the internal parameters of the camera, the relative distance and angle relationship between the active connection surface and the passive connection surface are deduced. The invention can not only measure the relative distance and relative rotation angle with the target object, but also identify different types of target objects and obstacles according to the requirements of the space cell robot, and the angle measurement range is also larger, which is suitable for any translation of the space cell robot Or external state perception under rotational motion.

Description

technical field [0001] The invention relates to a vision-based external state perception method of a space cell robot, which belongs to the field of self-reconfiguration of the space cell robot. Background technique [0002] In order to meet the needs of different space on-orbit tasks, the space cell robot can complete the configuration change of the robot group through self-reconfiguration, so as to achieve the purpose of expanding the function of the unit cell and expanding the working space of the robot. The method of self-reconfiguration of the space cell robot is through the autonomous connection between the unit cells. The key is to measure the relative distance and angle between the active and passive connection surfaces, and then issue corresponding control instructions. The external state perception of the space cell robot is a method to determine the relative position and attitude between the space cell robot and the external environment, targets and obstacles, and...

AI Technical Summary

Problems solved by technology

[0004] Although the traditional infrared sensor or ultrasonic sensor can realize the distance measurement, it cannot determine the type of the measurement target, and when measuring the relative rotation angle, it is necessary to install the corresponding infrared sensor on the measured object, and determine the angle by changing the range of the received signal. The detectable range of the angle is small; the distance measurement based on monocular vision generally adopts the triangulation method, but this method needs to initialize the translation of the camera before the measurement, and the distance cannot be calculated by the triangulation method for a monocular camera with pure rotational motion. And in the initialization process, if the translation distance is too small, it will lead to a large depth uncertainty and reduce the accuracy, while a large translation distance will cause the image of the target area to change greatly, which may invalidate the method, and the relative rotation angle cannot be measured.

Images