Robot system characterization: error modeling, identification, analysis, and minimization
This dissertation describes the development and application of a characterization methodology that improves the performance of robotic systems. To achieve accurate positioning, robot geometry must be precisely defined, both in new system design and in upgrading existing robots. This can be accomplis...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en |
| Published: |
Virginia Tech
2014
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10919/40223 http://scholar.lib.vt.edu/theses/available/etd-10262005-143525/ |
| Summary: | This dissertation describes the development and application of a characterization methodology that improves the performance of robotic systems. To achieve accurate positioning, robot geometry must be precisely defined, both in new system design and in upgrading existing robots. This can be accomplished by developing rigorous calibration methods to model, identify, analyze, and minimize errors in robot geometric parameters. Throughout this work, the geometric parameters that describe the kinematics of a given robot are treated as unknowns. The robot characterization process involves finding the optimal values of these parameters to best fit a set of measured or simulated positions of the robot end-effector.
In this dissertation, well-established robot kinematic link transformation techniques are first used to model the robotic manipulator system. Next, engineering knowledge of the robot system, its work environment and detailed component specifications are used to identify possible sources of error. This results in a list of error parameters and their range. A system sensitivity analysis is performed on these parameters to determine which have the greatest effect on system accuracy. To characterize an existing robot, experimental calibration data is gathered using a suitable measurement technique. Using this data, optimization of the previously isolated critical parameters is performed. The newly determined values of these parameters are implemented into the control system and performance is compared before and after the characterization process.
To design a new robot system, the isolated critical parameters are again found through optimization. In this case, however, the measured data 1s gathered by a simulation, with the error parameter values randomly generated each time. The performance of the system is analyzed after this exhaustive simulation. In both of the cases described, the characterization process is iterative.
The characterization process has been successfully applied to the design of a positioning system for a mobile, underwater nuclear-reactor-vessel-inspection robot. Also, the performance of an existing PUMA 562 industrial robot has been improved using this characterization procedure.
The advantages of this methodology over previous ones are that it can be applied to both new and existing robot systems and it is specifically aimed at meeting performance goals. A cost-performance tradeoff is accomplished by optimizing only for the critical parameters required to meet the specified performance objectives. === Ph. D. |
|---|