Grounded Reinforcement Learning for Visual Reasoning

TL;DR

ViGoRL is a novel vision-language reinforcement learning model that explicitly grounds each reasoning step in visual coordinates, significantly improving performance on diverse visual reasoning tasks.

Contribution

Introduces ViGoRL, a visually grounded RL framework that enhances spatial reasoning and visual attention in models, outperforming baselines across multiple benchmarks.

Findings

ViGoRL outperforms supervised and RL baselines on visual reasoning benchmarks.

Multi-turn RL with zoomed-in visual feedback improves localization of small GUI elements.

Grounding enhances exploration, subgoal setting, and visual verification behaviors.

Abstract

While reinforcement learning (RL) over chains of thought has significantly advanced language models in tasks such as mathematics and coding, visual reasoning introduces added complexity by requiring models to direct visual attention, interpret perceptual inputs, and ground abstract reasoning in spatial evidence. We introduce ViGoRL (Visually Grounded Reinforcement Learning), a vision-language model trained with RL to explicitly anchor each reasoning step to specific visual coordinates. Inspired by human visual decision-making, ViGoRL learns to produce spatially grounded reasoning traces, guiding visual attention to task-relevant regions at each step. When fine-grained exploration is required, our novel multi-turn RL framework enables the model to dynamically zoom into predicted coordinates as reasoning unfolds. Across a diverse set of visual reasoning benchmarks--including SAT-2 and BLINK for spatial reasoning, V*bench for visual search, and ScreenSpot and VisualWebArena for web-based grounding--ViGoRL consistently outperforms both supervised fine-tuning and conventional RL baselines that lack explicit grounding mechanisms. Incorporating multi-turn RL with zoomed-in visual feedback significantly improves ViGoRL's performance on localizing small GUI elements and visual search, achieving 86.4% on V*Bench. Additionally, we find that grounding amplifies other visual behaviors such as region exploration, grounded subgoal setting, and visual verification. Finally, human evaluations show that the model's visual references are not only spatially accurate but also helpful for understanding model reasoning steps. Our results show that visually grounded RL is a strong paradigm for imbuing models with general-purpose visual reasoning.