Robot programming by demonstration (PBD) system for task in which objectrequires contact with environment is built based on the controlling skill model. The skill isdescribed in three aspects: contact state classifier...Robot programming by demonstration (PBD) system for task in which objectrequires contact with environment is built based on the controlling skill model. The skill isdescribed in three aspects: contact state classifier, acquirement of contact states sequence andcontrolling transition between states. The classifier is developed with the support vector machineby using force sense. Sequence of states is obtained from the force signal of demonstration by theevent trigger. The velocity command of transition is achieved by linearization method. The PBDsystem is successfully built with robot controller with open architecture.展开更多
In this article,a new trajectory programming system that allows non-expert users to intuitively and efficiently program trajectories for robots is proposed.The system tracks a pen-shaped marker and obtains its positio...In this article,a new trajectory programming system that allows non-expert users to intuitively and efficiently program trajectories for robots is proposed.The system tracks a pen-shaped marker and obtains its position and orientation by processing the point cloud data of the workspace.A graphical user interface,which enables users to save and execute the acquired trajectory immediately after performing trajectory demonstration,is designed and developed for the system.The performance of the developed system is experimentally evaluated by using it to program trajectories for a UR5 robot.The results indicate that compared with traditional kinesthetic programming,the developed system has the potential of significantly reducing the ergonomic stress and workload of users.The system is developed based on the robot operating system,which facilitates its integration with different robot control systems.展开更多
To acquire human operation skill based on force sense, element contact form (ECF) is proposed to describe contact condition firstly. The skill is modeled as a sequence of discrete ECFs. Since different ECF has differe...To acquire human operation skill based on force sense, element contact form (ECF) is proposed to describe contact condition firstly. The skill is modeled as a sequence of discrete ECFs. Since different ECF has different force distribution, a support vector machine classifier is built to identify the contact conditions according to the force signal. Finally, the robot can obtain the skill from the human demonstration.展开更多
基金This project is supported by National Natural Science Foundation of China(No.6997S014) China 863 Robot Automatic Assembly Topic (No.863512972004).
文摘Robot programming by demonstration (PBD) system for task in which objectrequires contact with environment is built based on the controlling skill model. The skill isdescribed in three aspects: contact state classifier, acquirement of contact states sequence andcontrolling transition between states. The classifier is developed with the support vector machineby using force sense. Sequence of states is obtained from the force signal of demonstration by theevent trigger. The velocity command of transition is achieved by linearization method. The PBDsystem is successfully built with robot controller with open architecture.
基金supported by the Major Projects of Guangzhou City of China(Grant Nos.201907010012,201704030091 and 201607010041)the Guangdong Innovative and Entrepreneurial Research Team Program(Grant No.2014ZT05G132)+3 种基金Shenzhen Peacock Plan(Grant No.KQTD2015033117354154)the Major Projects of Guangdong Province of China(Grant No.2015B010919002)the Major Projects of Dongguan City of China(Grant No.2017215102008)the Nansha District International Science and Technology Cooperation Project of Guangzhou City of China(Grant No.2016GJ004).
文摘In this article,a new trajectory programming system that allows non-expert users to intuitively and efficiently program trajectories for robots is proposed.The system tracks a pen-shaped marker and obtains its position and orientation by processing the point cloud data of the workspace.A graphical user interface,which enables users to save and execute the acquired trajectory immediately after performing trajectory demonstration,is designed and developed for the system.The performance of the developed system is experimentally evaluated by using it to program trajectories for a UR5 robot.The results indicate that compared with traditional kinesthetic programming,the developed system has the potential of significantly reducing the ergonomic stress and workload of users.The system is developed based on the robot operating system,which facilitates its integration with different robot control systems.
文摘To acquire human operation skill based on force sense, element contact form (ECF) is proposed to describe contact condition firstly. The skill is modeled as a sequence of discrete ECFs. Since different ECF has different force distribution, a support vector machine classifier is built to identify the contact conditions according to the force signal. Finally, the robot can obtain the skill from the human demonstration.
