| With the development of collaborative robot technology,the collaborative work between humans and robots has become increasingly widespread.The manual operation has good intelligence and flexibility,while the robot automatic operation has excellent efficiency and precision.Human-robot collaboration can combine human intelligence and flexibility with the high-efficiency and high-precision of robots to complete a variety of complex tasks.Therefore,collaborative robots have been widely used in the manufacturing and service industries.To further improve the efficiency of the human-robot collaboration process and ensure the safety of human-robot-environment interaction,this paper proposes a compliant motion generation and generalization algorithm for collaborative robots based on dynamic feedforward.First,modeling,identification,and feedforward control of robot dynamics are carried out.Among them,the robot body dynamics is modeled by the Lagrangian method,and the dynamic parameters are identified by the two-step identification method after dynamic linearization and inertial parameter reorganization;the load dynamics is modeled,and the load parameter identification algorithm based on the virtual force/torque sensor is used to identify the load dynamic parameters;an improved Lu Gre friction model considering the influence of the load is established,and the friction parameter identification is carried out using the particle swarm optimization algorithm.Next,the robot variable impedance controllers based on dynamic feedforward are designed.Among them,the dynamic feedforward-based variable impedance control models in the joint space and task space are established,and the acceleration noise problem in the models is solved;a variety of impedance adaptive strategies are designed,and the stability of the robot’s motion is guaranteed by using the offline stability constraint method and the online stability adjustment method.finally,the robot’s compliant motion generation and generalization algorithms are studied.Among them,the robot motion generation methods such as motion planning and kinesthetic teaching are introduced,and the dynamic movement primitive algorithm is modeled,simulated,and analyzed;a compliant movement primitive algorithm is proposed,and related data preprocessing methods,the robot variable stiffness interface,and the structure of multi-dimensional compliant movement primitives are studied;a robot skill learning framework is designed to ensure the compliance and safety of robot task learning and generalization.To verify the proposed algorithms and strategies,two compliant motion experiments of the robot are designed.The compliant manipulation experiment of the robot proves the flexibility and efficiency of the kinesthetic teaching method and the compliance and safety of the joint space variable impedance control;the generalization experiment of robot compliant manipulation proves the intuitive compliant interaction ability of the task space variable impedance control and the generalization ability of the robot skill learning framework.Experimental results show that the algorithms and strategies proposed in this paper effectively improve the efficiency and safety of human-robot collaboration. |
