REVIEW 10 cited by
DemoGen: Synthetic Demonstration Generation for Data-Efficient Visuomotor Policy Learning
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Visuomotor policies have shown great promise in robotic manipulation but often require substantial amounts of human-collected data for effective performance. A key reason underlying the data demands is their limited spatial generalization capability, which necessitates extensive data collection across different object configurations. In this work, we present DemoGen, a low-cost, fully synthetic approach for automatic demonstration generation. Using only one human-collected demonstration per task, DemoGen generates spatially augmented demonstrations by adapting the demonstrated action trajectory to novel object configurations. Visual observations are synthesized by leveraging 3D point clouds as the modality and rearranging the subjects in the scene via 3D editing. Empirically, DemoGen significantly enhances policy performance across a diverse range of real-world manipulation tasks, showing its applicability even in challenging scenarios involving deformable objects, dexterous hand end-effectors, and bimanual platforms. Furthermore, DemoGen can be extended to enable additional out-of-distribution capabilities, including disturbance resistance and obstacle avoidance.
Forward citations
Cited by 10 Pith papers
-
DynamicManip: Enabling Dynamic Manipulation from a Single Static Demonstration
DynamicManip synthesizes diverse dynamic manipulation demonstrations from one static demonstration and uses stage-aware adaptive inference to improve success rates and reduce latency.
-
Static In, Dynamic Out: Counterfactual Action Augmentation for Moving Object Manipulation
SIDO morphs static demonstrations into counterfactual future-pose samples, training a goal-conditioned policy that, paired with a pose predictor, grasps objects whose motion was unseen during training.
-
Worlds in One Demo: A Synthetic Data Engine for Learning Open-World Mobile Manipulation
From one real demonstration, WANDA synthesizes diverse mobile-manipulation trajectories, reaching 54.8% average real-world task progress and zero-shot deployment on a morphologically different robot.
-
AffordGen: Generating Diverse Demonstrations for Generalizable Object Manipulation with Afford Correspondence
AffordGen synthesizes large-scale affordance-aware manipulation trajectories via keypoint correspondence on 3D meshes, enabling zero-shot visuomotor policies for unseen objects from few source demos.
-
DynaMimicGen: A Data Generation Framework for Robot Learning of Dynamic Tasks
DynaMimicGen generates large robot-training datasets from one or two demonstrations by adapting DMP-based trajectories in real time to moving object poses, improving downstream imitation-learning policies over MimicGen.
-
Learning in ImaginationLand: Omnidirectional Policies through 3D Generative Models (OP-Gen)
A robot policy trained on one real demonstration plus AI-generated 3D views succeeds from novel initial poses, including opposite-side starts, across six real manipulation tasks.
-
Constraint-Preserving Data Generation for Visuomotor Policy Learning
CP-Gen uses keypoint-trajectory constraints to turn a single expert demonstration into many geometry- and pose-varied robot demos, and policies trained on them transfer zero-shot to the real world.
-
TypeTele: Releasing Dexterity in Teleoperation by Dexterous Manipulation Types
A type-guided teleoperation system that selects predefined dexterous hand poses with a language model outperforms retargeting-based teleoperation on nine real-world tasks and improves imitation learning success.
-
DemoSpeedup: Accelerating Visuomotor Policies via Entropy-Guided Demonstration Acceleration
DemoSpeedup accelerates visuomotor policies by downsampling high-entropy segments of demonstrations, achieving roughly 2x faster execution with maintained or improved success rates.
-
ReinforceGen: Hybrid Skill Policies with Automated Data Generation and Reinforcement Learning
ReinforceGen uses imitation learning, RL fine-tuning of skill policies, and real-time pose replanning to reach over 80% success on five long-horizon Robosuite tasks from only 10 human demonstrations.
Discussion (0). Sign in to comment.