Bridging the Dynamic Perception Gap: Training-Free Draft Chain-of-Thought for Dynamic Multimodal Spatial Reasoning

TL;DR

This paper introduces D2R, a training-free framework that enhances multimodal large language models with dynamic visual drafts for improved spatial reasoning in evolving environments, demonstrated through a new maze navigation benchmark.

Contribution

The paper presents D2R, a novel training-free method integrating visual drafts with textual reasoning, and introduces GRASSLAND, a benchmark for dynamic spatial reasoning tasks.

Findings

D2R significantly improves performance on dynamic spatial reasoning tasks.

Augmenting textual chains with visual drafts outperforms traditional methods.

The approach works across diverse tasks without model fine-tuning.

Abstract

While chains-of-thought (CoT) have advanced complex reasoning in multimodal large language models (MLLMs), existing methods remain confined to text or static visual domains, often faltering in dynamic spatial reasoning tasks. To bridge this gap, we present GRASSLAND, a novel maze navigation benchmark designed to evaluate dynamic spatial reasoning. Our experiments show that augmenting textual reasoning chains with dynamic visual drafts, overlaid on input images, significantly outperforms conventional approaches, offering new insights into spatial reasoning in evolving environments. To generalize this capability, we propose D2R (Dynamic Draft-Augmented Reasoning), a training-free framework that seamlessly integrates textual CoT with corresponding visual drafts into MLLMs. Extensive evaluations demonstrate that D2R consistently enhances performance across diverse tasks, establishing a robust baseline for dynamic spatial reasoning without requiring model fine-tuning. Project is open at https://github.com/Cratileo/D2R.