Fast Uncertainty Quantification for Deep Object Pose Estimation

Deep learning-based object pose estimators are often unreliable and overconfident especially when the input image is outside the training domain, for instance, with sim2real transfer. Efficient and robust uncertainty quantification (UQ) in pose estimators is critically needed in many robotic tasks. In this work, we propose a simple, efficient, and plug-and-play UQ method for 6-DoF object pose estimation. We ensemble 2–3 pre-trained models with different neural network architectures and/or training data sources, and compute their average pair-wise disagreement against one another to obtain the uncertainty quantification. We propose four disagreement metrics, including a learned metric, and show that the average distance (ADD) is the best learning-free metric and it is only slightly worse than the learned metric, which requires labeled target data. Our method has several advantages compared to the prior art: 1) our method does not require any modification of the training process or the model inputs; and 2) it needs only one forward pass for each model. We evaluate the proposed UQ method on three tasks where our uncertainty quantification yields much stronger correlations with pose estimation errors than the baselines. Moreover, in a real robot grasping task, our method increases the grasping success rate from 35% to 90%. Video and code are available at https://sites.google.com/view/fastuq.

Authors

• Guanya Shi
• Yifeng Zhu
• Jonathan Tremblay
• Stan Birchfield
• Fabio Ramos 0001
• Anima Anandkumar
• Yuke Zhu

Keywords

• Measurement
• Training
• Uncertainty
• Computational modeling
• Pose estimation
• Neural networks
• Training data
• Uncertainty Quantification
• Human Pose Estimation
• Object Pose
• Neural Network
• Training Domain
• Spearman Correlation
• Rank Correlation
• Deep Models
• Point Cloud
• Bounding Box
• Confidence Score
• Ensemble Method
• Target Domain
• Heterogeneous Model
• Inference Time
• Prediction Uncertainty
• Domain Adaptation
• Robotic Applications
• Pose Prediction
• Deterministic Prediction
• Ground Truth Pose
• Image X
• Ensemble-based Methods
• Ketchup
• Ensemble Performance