Device and method for active grasping of manipulator based on multimodal fusion

Abstract

The invention provides a mechanical arm active grabbing device and method based on multi-model fusion. The mechanical arm active grabbing device based on multi-model fusion comprises a base (1), a mechanical arm (2), a laser radar (3), a binocular vision system (4) and a mechanical arm (5), wherein the laser radar (3) and one end of the mechanical arm (2) are respectively fixedly mounted on the base (1), and the binocular vision system (4) and the mechanical arm (5) are respectively fixedly mounted at the other end of the mechanical arm. The mechanical arm active grabbing method based on multi-model fusion comprises the following steps: step 1, sensing an object to be grabbed to obtain sensing information; step 2, positioning the object to be grabbed according to the sensing information toobtain positioning information; and step 3: grabbing the object to be grabbed according to the positioning information. The mechanical arm active grabbing device and method based on multi-model fusion fully consider the complex environment of space operation, effectively improve the capability of moving object grabbing, and have a wide application prospect.

Description

Technical field [0001] The present invention relates to the technical field of space robot positioning and grasping, in particular to a manipulator active grasping device and method based on multimodal fusion, in particular to a microgravity environment that combines binocular vision, lidar and tactile perception of a CMOS camera Robot positioning and active grasping technology. Background technique [0002] At present, the major countries in the world are accelerating the development of the aerospace field, and there are more and more life science experiments and space operations for space exploration. The development of traditional space activities relies on equipment preset instructions, direct operation by space station staff, or remote operation by ground staff. The lack of automatic real-time interaction and learning with the environment makes it difficult to achieve complex operations such as grabbing moving objects in a microgravity environment. task. The existing resea...

AI Technical Summary

Problems solved by technology

Traditional space activities rely on equipment preset instructions, direct operation by space station staff or remote operation by ground staff, lack of automatic real-time interaction and learning process with the environment, making it difficult to achieve complex operations such as grasping moving objects in a microgravity environment Task

The existing research on automatic grasping of moving objects in a microgravity environment mainly focuses on tactile perception combined with passive compliance mechanisms to overcome the impact of moving objects in the process of grasping to improve the success rate and reliability of grasping. There are few studies on the fusion of tactile, visual and other multi-modal information to realize the active grasping operation of manipulators. The correlation and complementarity between modal sensor information is of great significance to improve the grasping efficiency and robustness

Images