AI Paper Atlas

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

MobileVLA-R1: Reinforcing Vision-Language-Action for Mobile Robots

Ting Huang, Dongjian Li, Rui Yang, Zeyu Zhang, ZIDA YANG, Hao Tang

ECCV 2026 · Vision-language and multimodal › Robot action and navigation · approach: Vision-language-action foundation models · 16 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add GRPO reward stage after SFT with all three rewards (ablation)MobileVLA-R1 · R2R-CE Val-UnseenSR: 58.0 -> 68.3Improved
add GRPO reward stage after SFT with all three rewards (ablation)MobileVLA-R1 · R2R-CE Val-UnseenSPL: 53.2 -> 65.2Improved
add movement reward R_m to action+format rewards (ablation)MobileVLA-R1 · R2R-CE Val-UnseenSR: 65.2 -> 68.3Improved
add action reward R_action to movement+format rewards (ablation)MobileVLA-R1 · R2R-CE Val-UnseenSR: 63.4 -> 68.3Improved
add format reward R_format to movement+action rewards (ablation)MobileVLA-R1 · R2R-CE Val-UnseenSR: 64.5 -> 68.3Improved
add depth encoder to text and image encoder (ablation)MobileVLA-R1 · R2R-CE Val-UnseenSR: 62.5 -> 66.0Improved
add point cloud encoder on top of depth encoder (ablation)MobileVLA-R1 · R2R-CE Val-UnseenSR: 66.0 -> 67.2Improved
use full multimodal configuration instead of depth+point only (ablation)MobileVLA-R1 · R2R-CE Val-UnseenSR: 67.2 -> 68.3Improved
add depth and point encoders on RxR-CE (ablation)MobileVLA-R1 · RxR-CE Val-UnseenSR: 66.0 -> 71.5Improved
use MobileVLA-R1 instead of NaVILA initialization baselineMobileVLA-R1 · R2R-CE Val-UnseenSR: 54.0 -> 68.3Improved
use MobileVLA-R1 instead of MoREMobileVLA-R1 · QUARD (average success rate)success rate: 0.60 -> 0.73Improved
use MobileVLA-R1 instead of NaVILA in real-world complex corridorMobileVLA-R1 · Real-world Corridor (Complex)SR: 0.67 -> 0.86Improved

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