Get Started
Research questionHow can robots learn robust, generalist bimanual household manipulation from limited human demonstrations?Robustly coordinating both arms across varied household tasks requires broad behavioral coverage, but collecting high-quality human demonstrations at scale is costly and difficult. Limited demonstrations can leave policies brittle when task conditions or corrective actions differ from the training data.
Machine Learning
Multimodal Models
Robotics
Latest papersRecent research connected to this question, newest first.Scaling Bimanual Household Manipulation from 1,500 hours of Demonstrations to On-Policy CorrectionsThe source studies a vision-language-action model trained on 1,500 hours of diverse bimanual household manipulation demonstrations. It also examines scaling with additional expert data and post-training using DAgger correction data from real-time human interventions, with evidence limited to the task and data ranges evaluated.research paper · Sep 3, 2026
Related questions
How can dexterous robot hands generalize a manipulation skill from one demonstration to varied objects despite sim-to-real gaps?How can web videos provide relevant human demonstrations for long-tail dexterous robot manipulation?How can robots safely learn dynamic manipulation skills online despite sim-to-real mismatch?How can robots generalize residents’ packing preferences to new arrangements without continuous expert teleoperation?
Home
Topics
Search
Library