Paper
RoboAlign: Learning Test-Time Reasoning for Language-Action Alignment in Vision-Language-Action Models
Authors
Dongyoung Kim, Sumin Park, Woomin Song, Seungku Kim, Taeyoung Kim, Huiwon Jang, Jinwoo Shin, Jaehyung Kim, Younggyo Seo
Abstract
Improving embodied reasoning in multimodal-large-language models (MLLMs) is essential for building vision-language-action models (VLAs) on top of them to readily translate multimodal understanding into low-level actions. Accordingly, recent work has explored enhancing embodied reasoning in MLLMs through supervision of vision-question-answering type. However, these approaches have been reported to result in unstable VLA performance, often yielding only marginal or even negative gains. In this paper, we propose a more systematic MLLM training framework RoboAlign that reliably improves VLA performance. Our key idea is to sample action tokens via zero-shot natural language reasoning and refines this reasoning using reinforcement learning (RL) to improve action accuracy. As a result, RoboAlign bridges the modality gap between language and low-level actions in MLLMs, and facilitate knowledge transfer from MLLM to VLA. To validate the effectiveness of RoboAlign, we train VLAs by adding a diffusion-based action head on top of an MLLM backbone and evaluate them on major robotics benchmarks. Remarkably, by performing RL-based alignment after SFT using less than 1\% of the data, RoboAlign achieves performance improvements of 17.5\%, 18.9\%, and 106.6\% over SFT baselines on LIBERO, CALVIN, and real-world environments, respectively.
Metadata
Related papers
Fractal universe and quantum gravity made simple
Fabio Briscese, Gianluca Calcagni • 2026-03-25
POLY-SIM: Polyglot Speaker Identification with Missing Modality Grand Challenge 2026 Evaluation Plan
Marta Moscati, Muhammad Saad Saeed, Marina Zanoni, Mubashir Noman, Rohan Kuma... • 2026-03-25
LensWalk: Agentic Video Understanding by Planning How You See in Videos
Keliang Li, Yansong Li, Hongze Shen, Mengdi Liu, Hong Chang, Shiguang Shan • 2026-03-25
Orientation Reconstruction of Proteins using Coulomb Explosions
Tomas André, Alfredo Bellisario, Nicusor Timneanu, Carl Caleman • 2026-03-25
The role of spatial context and multitask learning in the detection of organic and conventional farming systems based on Sentinel-2 time series
Jan Hemmerling, Marcel Schwieder, Philippe Rufin, Leon-Friedrich Thomas, Mire... • 2026-03-25
Raw Data (Debug)
{
"raw_xml": "<entry>\n <id>http://arxiv.org/abs/2603.21341v1</id>\n <title>RoboAlign: Learning Test-Time Reasoning for Language-Action Alignment in Vision-Language-Action Models</title>\n <updated>2026-03-22T17:57:55Z</updated>\n <link href='https://arxiv.org/abs/2603.21341v1' rel='alternate' type='text/html'/>\n <link href='https://arxiv.org/pdf/2603.21341v1' rel='related' title='pdf' type='application/pdf'/>\n <summary>Improving embodied reasoning in multimodal-large-language models (MLLMs) is essential for building vision-language-action models (VLAs) on top of them to readily translate multimodal understanding into low-level actions. Accordingly, recent work has explored enhancing embodied reasoning in MLLMs through supervision of vision-question-answering type. However, these approaches have been reported to result in unstable VLA performance, often yielding only marginal or even negative gains. In this paper, we propose a more systematic MLLM training framework RoboAlign that reliably improves VLA performance. Our key idea is to sample action tokens via zero-shot natural language reasoning and refines this reasoning using reinforcement learning (RL) to improve action accuracy. As a result, RoboAlign bridges the modality gap between language and low-level actions in MLLMs, and facilitate knowledge transfer from MLLM to VLA. To validate the effectiveness of RoboAlign, we train VLAs by adding a diffusion-based action head on top of an MLLM backbone and evaluate them on major robotics benchmarks. Remarkably, by performing RL-based alignment after SFT using less than 1\\% of the data, RoboAlign achieves performance improvements of 17.5\\%, 18.9\\%, and 106.6\\% over SFT baselines on LIBERO, CALVIN, and real-world environments, respectively.</summary>\n <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n <published>2026-03-22T17:57:55Z</published>\n <arxiv:comment>15 pages, 7 figures, 9 Tables</arxiv:comment>\n <arxiv:primary_category term='cs.AI'/>\n <author>\n <name>Dongyoung Kim</name>\n </author>\n <author>\n <name>Sumin Park</name>\n </author>\n <author>\n <name>Woomin Song</name>\n </author>\n <author>\n <name>Seungku Kim</name>\n </author>\n <author>\n <name>Taeyoung Kim</name>\n </author>\n <author>\n <name>Huiwon Jang</name>\n </author>\n <author>\n <name>Jinwoo Shin</name>\n </author>\n <author>\n <name>Jaehyung Kim</name>\n </author>\n <author>\n <name>Younggyo Seo</name>\n </author>\n </entry>"
}