Unified Probabilistic Model for Robot Calibration Fusing Kinematic, Sensor, and Manufacturing Priors
Abstract
Robot calibration methods that rely on external sensor measurements may face challenges with accuracy and robustness, for example, in the presence of sensor noise, occlusions, or geometric singularities. A disclosed technology can utilize a unified probabilistic model, which may be implemented as a factor graph, to fuse heterogeneous data sources. This model can integrate data sources, such as high-frequency robot control inputs like encoder trajectories and manufacturing specifications like computer-aided design tolerances, as probabilistic priors with sensor measurements. A reliability gating mechanism may help manage geometric degeneracies, for instance, by dynamically weighting data sources based on observability. This approach may improve calibration accuracy and robustness, promote the physical plausibility of results, and provide more stable parameter estimation, for example, when sensor observability is degraded.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Kim, Taemin, "Unified Probabilistic Model for Robot Calibration Fusing Kinematic, Sensor, and Manufacturing Priors", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10065