arXiv preprint ·
Learning Keypoint Constraints from Visual Demonstrations for Safe Motion Planning
- Published
- Source
- arXiv preprint, posted before peer review
- Original
- Read it at arXiv
- Topics
Abstract
We present Keypoint Constraint Learning (KCL), a framework for learning task-relevant motion constraints from visual demonstrations and for planning safely with the extracted constraints. To make constraint learning tractable from RGB-D observations, we represent demonstrations using sparse 3D object-centric keypoints and semantic mask centroids extracted through segmentation, visual feature clustering, and point tracking. In this reduced geometric space, we learn keypoint dynamics and infer distance- and orientation-based constraints that make the expert trajectories feasible and approximately locally optimal by enforcing KKT optimality conditions in the reduced state space. We then plan with the learned constraints under uncertainty, using robustness margins derived from accumulated prediction error, and provide a theoretical analysis of how keypoint perception error and learned dynamics error affect the recovered constraints. We evaluate KCL on visual demonstrations from navigation to rigid/deformable manipulation, showing that it recovers multi-stage geometric constraints and outperforms baselines in downstream safe planning.
The authors’ abstract, as arXiv lists it. arXiv’s listings, abstracts included, are free for anyone to reuse.
Read it on ReadSearch
Each opens after you create an account, and brings you back to this paper.
More in Robotics
An Optimal Algorithm for Changing from Longitudinal to Latitudinal Formation of Autonomous Aircraft Squadrons
NavAble: A Large-Scale Dataset and Synthetic Data Generation Pipeline for Blind Navigation
YUBI-Real2Sim: 3DGS Scene Verification for Reliable Policy Evaluation
Probabilistic Physics Digital Twin for Human-Robot Interaction
Box-WAM: Box Trajectories as Compact State Representations for World Action Models
Get the few papers worth your time in Robotics every morning.
ReadSearch reads the new papers in your field each day and brings you the few worth reading, each with its key finding in a sentence.