A new closed-loop output error method for parameter identification of robot dynamics (1009.4556v1)

Abstract: Off-line robot dynamic identification methods are mostly based on the use of the inverse dynamic model, which is linear with respect to the dynamic parameters. This model is sampled while the robot is tracking reference trajectories that excite the system dynamics. This allows using linear least-squares techniques to estimate the parameters. The efficiency of this method has been proved through the experimental identification of many prototypes and industrial robots. However, this method requires the joint force/torque and position measurements and the estimate of the joint velocity and acceleration, through the bandpass filtering of the joint position at high sampling rates. The proposed new method requires only the joint force/torque measurement. It is a closed-loop output error method where the usual joint position output is replaced by the joint force/torque. It is based on a closed-loop simulation of the robot using the direct dynamic model, the same structure of the control law, and the same reference trajectory for both the actual and the simulated robot. The optimal parameters minimize the 2-norm of the error between the actual force/torque and the simulated force/torque. This is a non-linear least-squares problem which is dramatically simplified using the inverse dynamic model to obtain an analytical expression of the simulated force/torque, linear in the parameters. A validation experiment on a 2 degree-of-freedom direct drive robot shows that the new method is efficient.

• The paper introduces the Direct and Inverse Dynamic Identification Model (DIDIM), a new closed-loop output error method for robot dynamics identification requiring only joint force/torque data.
• Experimental validation shows the DIDIM method is robust against initialization, closed-loop simulation errors, and low data sampling frequencies.
• The method simplifies the identification process by eliminating the need for velocity and acceleration derivative estimates and could inform future control algorithm development.

An Evaluation of the Closed-Loop Output Error Method for Robot Dynamics Parameter Identification

The paper by Gautier, Janot, and Vandanjon introduces a novel approach for identifying the dynamic parameters of robots using a closed-loop output error method. Traditionally, offline identification of robot dynamics relies on the inverse dynamic model (IDM), dependent on joint force/torque and position measurements, along with estimates of joint velocity and acceleration, necessitating high-frequency sampling. The proposed method simplifies this by requiring solely the joint force/torque measurements, circumventing the need for joint position measurements, and effectively employing both the inverse and direct dynamic models for identification, hereby termed the DIDIM (Direct and Inverse Dynamic Identification Model) method.

Overview of Methodology

The DIDIM approach redefines the output error method by substituting joint force/torque for joint position. The identification involves simulating a closed-loop system using the direct dynamic model with the same control architecture and reference trajectory as the actual robotic system, minimizing the 2-norm of the error between the actual and simulated forces/torques. This strategy formulates a non-linear least-squares problem that becomes streamlined via analytical expressions derived from the inverse dynamic model. By sidestepping the necessity for joint position data, the method qualifies as robust against inaccuracies in the velocity and acceleration estimations needed in traditional IDM approaches.

Experimental Validation

The paper presents evidence from experiments on a 2 degree-of-freedom direct drive robot, underlining the robustness and efficiency of the DIDIM approach. The method not only rivals traditional IDIM outputs when optimal initial conditions and filtering are employed, but it also demonstrates robustness regarding:

• Initialization: It exhibits independence from precise parameter initialization, utilizing a regular matrix for the inertia parameters to start the integration process effectively.
• Closed-loop simulation errors: DIDIM is resilient to differences in tuning between the actual and simulated systems, provided the control law structure is preserved.
• Data sampling frequency: In contrast to IDIM, which relies heavily on high-frequency joint position data, DIDIM maintains accuracy even at lower sampling rates and without data filtering.

Implications and Speculative Developments

Practically, the DIDIM method reduces identification complexity by eliminating the need for derivative estimates of joint velocities and accelerations. This reduces errors due to phase distortions induced by filtering at high sampling rates. Theoretically, this approach could inform future developments in control algorithms and dynamic simulations where accurate parameter identification remains crucial.

Looking forward, the implementation of DIDIM on more complex robot systems, such as those with six degrees of freedom, could validate its scalability and adaptability across different robotic platforms. This could further influence how parameter identification is approached in scenarios involving dynamic control and simulation fidelity, especially in haptic and collaborative robotics requiring precise force control.

In summary, this paper presents a beneficial trajectory for robotics research, potentially simplifying dynamic parameter identification and enhancing control systems through precise model validations. However, it is key to ensure that the initial conditions and control structures are accurately simulated to fully leverage the advantages of the DIDIM method in diverse settings.