AI Paper Atlas

Research map / Vision-language and multimodal / Robot action and navigation

NestedVLA: Learning to Consolidate and Generate Skills for Vision-Language-Action Model

Guodong DU, Wanyu Lin

NeurIPS 2026 · Vision-language and multimodal › Robot action and navigation · approach: Vision-language-action foundation models

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes NestedVLA, a paradigm for consolidating robot skills learned across heterogeneous tasks, environments, and embodiments in a vision-language-action model, as an alternative to joint multi-task training and model merging.

Problem

Consolidating skills learned across heterogeneous robotic tasks, environments, and embodiments into one VLA model is hard: data is fragmented, joint multi-task training suffers interference, and model merging causes parameter conflicts or needs architectural changes.

Approach

Advocates a paradigm in which skills are learned and consolidated as nested modules within a VLA, rather than via joint training or parameter merging; details unavailable from the abstract provided.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
generate action-expert update with a hypernetwork instead of static or dynamic mergingNestedVLA · MetaWorld, ManiSkill, CALVIN averagesuccess rate: 25.6 -> 66.7Improved
add per-task residual storage to merging (dynamic merging) at 2.09x-2.89x storagemodel merging baselines · average successsuccess rate: 14.8 (Averaging) -> 22.6-25.6Improved
add sign-conflict resolution or singular-vector compression to averagingstatic merging (TIES, WUDI, TSV) · average successsuccess rate: 14.8 -> 13.1 / 11.8 / 13.4Hurt
add per-task output heads (multi-head path) to hypernetwork (ablation)NestedVLA hypernetwork · MetaWorld/ManiSkill/CALVINin-mask standardized loss: MW 1.33 -> 1.08Improved
add per-task patch embeddings (ablation)NestedVLA hypernetwork · MetaWorld/ManiSkill/CALVINin-mask standardized loss: MW 1.02 (without) vs 1.08 (default)No clear effect
add shared-FC LoRA path (ablation)NestedVLA hypernetwork · MetaWorld/ManiSkill/CALVINin-mask standardized loss: MW 0.91 (without) vs 1.08 (default)Hurt
narrow hypernetwork hidden width from 2048 to 1024 (ablation)NestedVLA hypernetwork · MetaWorldin-mask standardized loss: 1.08 -> 1.43Hurt
use drop-and-rescale instead of 1-bit quantization to compress task vectors (ablation)NestedVLA skill compression · per-task fidelitysuccess penalty vs FlowSDE: -71 to -81 pp -> -3 to -5 ppImproved
add rank-r SVD residual to drop-and-rescale (ablation)NestedVLA skill compression · per-task fidelityreconstruction error: at most +0.005 fidelityNo clear effect
disable out-of-mask penalty (w_out = 0) (ablation)NestedVLA training loss · average successsuccess rate: +2.4 ppImproved
add magnitude weighting or absolute loss space alone (ablation)NestedVLA training loss · average successsuccess rateHurt
raise keep ratio K from 10% to 20% (ablation)skill compression ideal · MetaWorldsuccess rate: 48.4 -> 60.2Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy