Temporally Consistent Object 6D Pose Estimation for Robot Control
For roboticists needing temporally consistent 6D object pose estimates for stable vision-based control, this work addresses a key limitation of off-the-shelf estimators.
The paper develops a factor graph approach to enforce temporal consistency in single-view RGB object pose estimation, improving stability for robot control. The method achieves significant improvements on standardized benchmarks and demonstrates stable feedback control in a real robot tracking task.
Single-view RGB object pose estimators have reached a level of precision and efficiency that makes them good candidates for vision-based robot control. However, off-the-shelf methods lack temporal consistency and robustness that are mandatory for a stable feedback control. In this work, we develop a factor graph approach to enforce temporal consistency of the object pose estimates. In particular, the proposed approach: (i) incorporates object motion models, (ii) explicitly estimates the object pose measurement uncertainty, and (iii) integrates the above two components in an online optimization-based estimator. We demonstrate that with appropriate outlier rejection and smoothing using the proposed factor graph approach, we can significantly improve the results on standardized pose estimation benchmarks. We experimentally validate the stability of the proposed approach for a feedback-based robot control task in which the object is tracked by the camera attached to a torque controlled manipulator.