Robot stiffness identification depends not only on the measurements collected but also on the robot configurations used to collect them. For a 7-degree-of-freedom (7-DoF) collaborative robot, selecting useful measurement poses presents a particular challenge: the extra degree of freedom makes the manipulator redundant, so different joint configurations can place the robot in comparable task-space positions while producing different conditions for parameter identification.
A configuration-selection method for elastostatic calibration addresses this issue by examining how individual joints influence an evaluation criterion called the inverse condition number. The method is intended to help identify joint stiffness more accurately by choosing measurement configurations from the robot’s joint space rather than relying on configurations selected without a systematic assessment.
Why measurement configuration matters in robot stiffness identification
Elastostatic calibration concerns a robot’s response to applied loads under static conditions. In stiffness identification, measurements are used to estimate the stiffness associated with the robot’s joints. The usefulness of those measurements depends in part on whether the chosen configurations allow the relevant stiffness parameters to be distinguished.
If a set of configurations provides weak or poorly differentiated information about the parameters, identification can be less reliable. Configuration selection is therefore part of the calibration problem, not simply a preliminary step. This issue becomes more involved for redundant manipulators, where more than one joint arrangement may be available for a given end-effector position.
For a 7-DoF collaborative robot, the redundant degree of freedom creates additional options in joint space. Those options make it possible to choose among configurations, but they also mean that the effect of joint arrangement needs to be considered explicitly when preparing measurements. A method that accounts for redundancy can help avoid treating all candidate configurations as equivalent.
Separating the influence of individual joints
The proposed approach uses an influencing factor separation method to study the relationship between joint configuration and the inverse condition number. Rather than considering the configuration only as a single combined choice, it analyzes the independent influence of each joint on the selected evaluation criterion.
To do this, the method applies an orthogonal design experiment and analysis of variance. In practical terms, the experimental design provides a structured way to examine joint-factor combinations, while the analysis of variance is used to assess the contribution of the factors being studied. This helps identify which joint choices influence the criterion and supports selection of configurations within the robot’s joint space.
The inverse condition number serves as the measure for comparing candidate configurations in this method. It provides a basis for assessing configuration suitability for stiffness identification. The reported results indicate that configurations with a large inverse condition number produced more accurate identification than configurations associated with medium or small values in the study’s comparisons.
What the 7-DoF robot study evaluated
The method was applied to a SHIR5-III 7-DoF collaborative robot. Static compliance simulations were performed to identify the robot’s joint stiffness. The identification results obtained using configurations with large, medium, and small inverse condition numbers were then compared.
This comparison is important because it evaluates the configuration-selection approach against alternatives within the same stated framework. The reported outcome was that using configurations with a large inverse condition number could essentially improve identification accuracy. The finding supports the use of the criterion as a guide for choosing measurement configurations for this stiffness-identification task.
The evaluation described is simulation-based. It should therefore be understood as evidence from static compliance simulations for the robot and method examined, rather than as a claim that the same accuracy improvement has been demonstrated for every collaborative robot, calibration setup, or operating environment.
Practical considerations for calibration planning
For engineers and system integrators planning stiffness calibration, the work highlights a useful distinction: collecting more measurements is not necessarily the same as collecting more informative measurements. The configuration of the manipulator can affect the quality of the identification process, so candidate poses should be assessed rather than chosen solely for convenience.
A practical planning workflow consistent with the method would include:
- Define the identification task. Establish that the objective is to identify joint stiffness through static compliance measurements.
- Account for redundancy. Treat different joint-space configurations as meaningful alternatives, even when they correspond to similar task-space positions.
- Evaluate joint effects systematically. Use a structured approach to understand how individual joints influence the selected configuration criterion.
- Compare candidate configurations. Use the inverse condition number to guide selection, with the study reporting better identification results for configurations with large values than for those with medium or small values.
These steps do not replace the need to validate a calibration procedure for a particular robot and measurement setup. They emphasize that configuration choice can be incorporated into the planning process and evaluated with a defined criterion.
Conclusion
For redundant 7-DoF collaborative robots, measurement configuration is a consequential part of elastostatic calibration and joint stiffness identification. Separating the influence of individual joints through orthogonal design and analysis of variance offers a structured way to select configurations from joint space. In simulations on the SHIR5-III, configurations with a large inverse condition number yielded more accurate stiffness-identification results than those with medium or small values. For automation teams, the practical implication is to assess the information value of candidate configurations as part of calibration planning, while keeping the study’s simulation scope in view.
Access Official Technical PDF