QDepth-VLA: Quantized Depth Prediction as Auxiliary Supervision for Vision-Language-Action Models
This addresses the need for better 3D understanding in VLA models for fine-grained manipulation, but it is incremental as it builds on existing frameworks with an added depth component.
The paper tackled the problem of enhancing spatial perception in Vision-Language-Action models for manipulation tasks by introducing an auxiliary depth prediction task, resulting in strong spatial reasoning and competitive performance on benchmarks.
Spatial perception and reasoning are crucial for Vision-Language-Action (VLA) models to accomplish fine-grained manipulation tasks. However, existing approaches often lack the ability to understand and reason over the essential 3D structures necessary for precise control. To address this limitation, we propose QDepth-VLA, a general framework that augments VLA models with an auxiliary depth prediction task. A dedicated depth expert is designed to predict quantized latent tokens of depth maps obtained from a VQ-VAE encoder, enabling the model to learn depth-aware representations that capture critical geometric cues. Experimental results on the simulation benchmarks and real-world tasks demonstrate that QDepth-VLA yields strong spatial reasoning and competitive performance on manipulation tasks.