Pith. sign in

REVIEW 5 cited by

Generalization in Dexterous Manipulation via Geometry-Aware Multi-Task Learning

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

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

arxiv 2111.03062 v1 pith:K4JLMF4V submitted 2021-11-04 cs.RO cs.AIcs.CVcs.LGcs.SYeess.SY

Wenlong Huang , Igor Mordatch , Pieter Abbeel , Deepak Pathak This is my paper
classification cs.ROcs.AIcs.CVcs.LGcs.SYeess.SY
keywords learningobjectsmanipulationmulti-taskobjectpoliciesdexterousgeneralist
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Dexterous manipulation of arbitrary objects, a fundamental daily task for humans, has been a grand challenge for autonomous robotic systems. Although data-driven approaches using reinforcement learning can develop specialist policies that discover behaviors to control a single object, they often exhibit poor generalization to unseen ones. In this work, we show that policies learned by existing reinforcement learning algorithms can in fact be generalist when combined with multi-task learning and a well-chosen object representation. We show that a single generalist policy can perform in-hand manipulation of over 100 geometrically-diverse real-world objects and generalize to new objects with unseen shape or size. Interestingly, we find that multi-task learning with object point cloud representations not only generalizes better but even outperforms the single-object specialist policies on both training as well as held-out test objects. Video results at https://huangwl18.github.io/geometry-dex

Discussion (0). Sign in to comment.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Revisiting Mixture Policies in Entropy-Regularized Actor-Critic

    cs.LG 2026-05 unverdicted novelty 7.0 of 10

    A new marginalized reparameterization estimator allows low-variance training of mixture policies in entropy-regularized actor-critic algorithms, matching or exceeding Gaussian policy performance in several continuous ...

  2. Rotation-Aware Point-Cloud Embeddings for Vision-Based In-Hand Reorientation

    cs.RO 2026-06 unverdicted novelty 6.0 of 10

    Learns rotation-aware point-cloud embeddings calibrated to SO(3) geodesic error, enabling model-free RL for vision-based in-hand reorientation without pose or flow inputs.

  3. Sparse2Act: Learning Action-Aligned Sparse 3D Representations for Cross-Domain Robot Manipulation

    cs.RO 2026-06 unverdicted novelty 5.0 of 10

    Sparse2Act pretrains sparse 3D encoders via masked action-alignment supervision, yielding reusable representations that reach 86.9% success on LIBERO-10 and enable cross-domain transfer.

  4. When Dynamics Shift, Robust Task Inference Wins: Offline Imitation Learning with Behavior Foundation Models Revisited

    cs.LG 2026-05 unverdicted novelty 5.0 of 10

    Robust minimax task inference in BFMs achieves dynamics-shift robustness from nominal offline data alone and outperforms standard baselines.

  5. EaDex: A Cross-Embodiment Dexterous Manipulation Framework from Low-Cost Demonstrations

    cs.RO 2026-06 unverdicted novelty 4.0 of 10

    EaDex combines single-camera RGB-D capture, MANO retargeting, and dynamic demonstration annealing to achieve 55.3% relative improvement over baseline on nine cross-embodiment dexterous object-opening tasks across three hands.

Pith tools