TY - GEN
T1 - Symmetry and Uncertainty-Aware Object SLAM for 6DoF Object Pose Estimation
AU - Merrill, Nathaniel
AU - Guo, Yuliang
AU - Zuo, Xingxing
AU - Huang, Xinyu
AU - Leutenegger, Stefan
AU - Peng, Xi
AU - Ren, Liu
AU - Huang, Guoquan
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - We propose a keypoint-based object-level SLAM framework that can provide globally consistent 6DoF pose estimates for symmetric and asymmetric objects alike. To the best of our knowledge, our system is among the first to utilize the camera pose information from SLAM to provide prior knowledge for tracking keypoints on symmetric objects - ensuring that new measurements are consistent with the current 3D scene. Moreover, our semantic key-point network is trained to predict the Gaussian covariance for the keypoints that captures the true error of the prediction, and thus is not only useful as a weight for the residuals in the system's optimization problems, but also as a means to detect harmful statistical outliers without choosing a manual threshold. Experiments show that our method provides competitive performance to the state of the art in 6DoF object pose estimation, and at a real-time speed. Our code, pre-trained models, and keypoint labels are available https://github.com/rpng/suo_slam.
AB - We propose a keypoint-based object-level SLAM framework that can provide globally consistent 6DoF pose estimates for symmetric and asymmetric objects alike. To the best of our knowledge, our system is among the first to utilize the camera pose information from SLAM to provide prior knowledge for tracking keypoints on symmetric objects - ensuring that new measurements are consistent with the current 3D scene. Moreover, our semantic key-point network is trained to predict the Gaussian covariance for the keypoints that captures the true error of the prediction, and thus is not only useful as a weight for the residuals in the system's optimization problems, but also as a means to detect harmful statistical outliers without choosing a manual threshold. Experiments show that our method provides competitive performance to the state of the art in 6DoF object pose estimation, and at a real-time speed. Our code, pre-trained models, and keypoint labels are available https://github.com/rpng/suo_slam.
KW - 3D from multi-view and sensors
KW - Motion and tracking
KW - Pose estimation and tracking
KW - Robot vision
KW - Scene analysis and understanding
KW - Vision applications and systems
UR - https://www.scopus.com/pages/publications/85142498957
U2 - 10.1109/CVPR52688.2022.01448
DO - 10.1109/CVPR52688.2022.01448
M3 - Conference contribution
AN - SCOPUS:85142498957
T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
SP - 14881
EP - 14890
BT - Proceedings - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022
PB - IEEE Computer Society
T2 - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022
Y2 - 19 June 2022 through 24 June 2022
ER -