Hegao Cai

Generation of closed-form inverse kinematics for reconfigurable robots

For reconfigurable robots, the automatic generation of inverse kinematics is a key problem, because such robots may assume various configurations. In this paper, the screw and product-of-exponentials (POE) formula are used to model the kinematics of reconfigurable robots. The POE formula can be converted to canonical subproblems through decomposition and adjoint transformation. Three classes and 28 types of subproblems containing geometric or algebraic solutions are identified and solved, which can be reused in different configurations. A generalized, decomposable, and reusable approach for close-form inverse kinematics of reconfigurable robots is developed based on POE and subproblems. The effectiveness of this method is shown in an example.

Biological Characteristics Analysis and Mechanical Jumping Leg Design for Frog Robot

This paper describes the biological observation and kinematic analysis of frog jump modality from three phases, and the use of these data to combine and simplify the frogs movement. The dynamic simulation based on these data provides the ground reaction forces which aid in robot design. A kind of hindlimb model is put forward, and corresponding jumping leg robot is designed according to trajectory and function of the limbs main joints. The robot has 5-bar spring/linkage leg and actuated by a DC motor. Its jump ability that studied and compared with frog is proved to be good efficiency. The results of analysis, simplified model and mechanical leg will be employed to develop and control a whole robot capable of mimicking the jumping behavior of the frog.

Passivity based control of flexible joint robot with dynamic friction compensation

As the friction has a great effect on the performance of low speed and low stiffness impedance control of flexible joint robot, a modified Dahl friction model is adopted, and a friction observer is proposed to compensate the friction based on interconnection and damping assignment passivity based control (IDA-PBC) method. The dynamics friction compensation not only improves the position and force tracking performance, but also assures the stability of the system. Finally, experiments are carried out on a 6-DOF anthropomorphic manipulator to verify the compensation. The results show that position accuracy as well as torque vibration of the flexible joint system are evidently improved in the position control. In addition, the force tracking ability of impedance control is also improved by the friction compensation.

Singularity Analysis of a 6-PRRS Parallel Robot Based on Geometrical Method

Three types of singularities have been analyzed for a 6-PRRS parallel robot in this paper, i.e., the forward, inverse, and combined singularity. Conventional analysis method of a parallel robot is based on the whole Jacobian matrix, so it needs complex process and computation. The singularity analysis in this paper focuses on the geometrical character of one motion chain, and every chain is processed by the same method, so the complexity of singularity analysis is significantly reduced. Using the product-of-exponential formula, the kinematics equations possess algebraic character so that the inverse singularity analysis algorithms can be readily and systematically formulated. By analyzing the instantaneous mobility of the moving platform, the necessary and sufficient geometrical conditions for the forward singularity configurations are identified. It has shown that these methods can be used for singularity analysis of various parallel robots

Interconnection and damping assignment passivity-based control for flexible joint robot

This paper presents an interconnection and damping assignment passivity based control (IDA-PBC) for the position and impedance control of flexible joint robot. In considering of the joint damping and with some simplification of friction, a flexible joint robot system is descripted as an port-controlled Hamiltonian system with dissipation(PCHD). Then, giving a desired Hamiltonian function which both shapes potential and kinetic energies, the control laws of position control and impedance control is derived by the algebraic IDA-PBC method, which preserves the PCHD structure of the close-loop system. Comparing with the traditional passivity based control (PBC), the proposed method gives a new explanation of the control of flexible joint robot by introducing the self-damping and cross-damping terms, which make the gain selecting procedure more intuitive and comfortable. Simulation and experiments are done to verify the proposed controller, and the results shows a promising performance both in position and impedance control.

Dynamical Research Based on a New Type of Reconfigurable Robot

Currently, the reconfigurable robot has become a hot topic for its diverse configurations. In this thesis, a new type of reconfigurable robot system is proposed, whose main modules are cylinders, which makes its structure quite compact and dexterous. Taking this kind of robot as an example, some new methods for robot module modeling and reconfiguration modeling are figured out, according to which a reconfigurable robot of any configuration can set up a dynamical model for itself automatically. By analyzing the velocities, forces and torques of each module, the dynamical functions of the robot are established, using the Newton-Euler method. Finally the six degree of freedom PUMA robot is used to carry out the experimental simulation to test the reliability and applicability of this method proposed in this thesis.

Architecture of Multi-Agent-Based Multi-Operator Multi-Mobile-Manipulator Teleoperation System

Being characteristics of motion, coordination, distribution, non-determinism etc, multi-operator multi-mobile-manipulator teleoperation system has difficulties in controlling and implementation. Multi-Agent-Based architecture of the system can distinctly describe the information interactive relationship between components of the system, and optimize intelligence distribution between human and robot, so it helps to reduce difficulties in controlling and implementation of the system. After the physical structure of the system was analyzed, the system was divided into several sub-systems with given function, and then a Multi-Agent-Based layered architecture with shared control mode of the system was proposed and the function of each Agent was planned. Multi-Agents mixed communication structure was designed by both of blackboard model and Peer-to-Peer model. According to characteristics of the system, a kind of general mixed architecture for Agent was proposed, and an example of implementation method of Agent was given, then the validity of the mixed architecture was verified by designing experiment.

Transparency Oriented Virtual Coupling Design: New Approach and Application to a Novel Admittance Haptic Device

A novel admittance haptic device which has low inertia and high speed to guarantee the high force bandwidth for the haptic interaction system is presented. Friction wheel is used to implement a rapid adjustment of the velocity loop. According to the admittance character of the haptic interface, PI velocity feedback control is utilized in the haptic interaction system. By introducing the virtual coupling into the system, Llewellyns stability criteria in circuit theory is used to derive the constraint conditions to guarantee the systems absolute stability. Furthermore the parameters of the virtual coupling are optimized oriented to high transparency of the system. Man-machine haptic interaction system is established with the novel admittance interface. The experimental results indicate that optimized virtual coupling guarantees the large range stability, as well as improving the transparency effectively.

Simulation Research on Reconfiguration Locomotion Planning for a 3-D Self-reconfigurable Robot

Self-reconfigurable robot consists of numerous modules. It has various configurations and multiform locomotion modes. How to describe its configurations and how to plan its locomotion are difficult problems. To solve these problems, a novel description of modules and their surroundings was presented and modular locomotion rule base was established. Based on the rule base, a reconfiguration locomotion planning algorithm combined heuristic search with limited-horizon search was proposed for a self-reconfigurable robot HitMSR. In heuristic search second selecting method was adopted to minimize search space. Simulation results indicated that the algorithm can generate reconfiguration locomotion series automatically and efficiently in complex environments for 3-D self-reconfigurable robot with any initial configuration

A Skill-Based Autonomous Control Method for Robotic Teleoperation

Robot autonomous control is emerging as a powerful method for robotic teleoperation. Human skill researches provide a new prospect for robot autonomous control, by transferring the skill information into the control of robot system. The paper presents a novel skill-based autonomous control method and also an implementation of skill-based controller including optimal filter, skill learner and force-controller. The experiments of robotic peg-in-hole task based human skills are conducted and prove the effectiveness of the control method.