Binocular Vision-based Care Robot Hold To Take The Realization Of The Function

With the arrival of the aging society, in order to take care of the elderly, patients and inconvenient people, the nursing robot was developed and it became the focus field of theory research and engineering experimental. The nursing robots is a integrated system of environmental awareness, dynamic planning and decision-making, and behavior control, involving mechanical, sensors, electronics, automation, artificial intelligence, computer and multi-disciplinary research.The Ingenious Company and Nanjing University of Science and Technology jointly developed MT-Bear robot experimental platform, research the kinematics and trajectory planning, dynamics and control of the dual-arm of it. Based on the binocular vision, we study the technology of image processing, target recognition and target location. The typical nursing, ambulance operations and the basic movements can be carried out.MT-Bear can take hold of an object smoothly and supplely. This major work includes the following:(1) The MT-Bear robot system is analyzed, including its humanoid dual-arm, binocular vision, a crawler type walking structure, and compare it with previous nursing robots. Use the D-H method to model the kinematics of MT-Bear robot arms, research on the kinematics forward and reverse solution of the three degrees of freedom manipulator and the motion trajectory of path planning, analyze quality of different planning path.(2) Use the Lagrange method to model the dynamics of MT-Bear manipulator, the PD control and variable structure control was used for manipulator control. Use Matlab to simulate the process, analyze the two kind of control and point out the advantages and disadvantages, and puts forward the improvement.(3) Based on binocular vision of the MT-Bear robot platform, do some research on how to recognize the target and analyze the advantages and disadvantages of each method, put forward suitable for MT-Bear robot target identification methods, Lay the foundation for the realization of take hold of target objects.(4) Briefly introduce the Webots simulation platform. Build the model in the robot 3D simulation and use control algorithm to realize the robot path planning and target hold by the three-dimensional simulation. Ported to MT-Bear to complete the simulation of take hold of the target, finally the two methods are compared.