Compressor-VLA: Instruction-Guided Visual Token Compression for Efficient Robotic Manipulation

A new study on Vision-Language-Action (VLA) models highlights the importance of action chunk length, termed horizon, in robotic manipulation. The research reveals a trade-off between longer horizons, which enhance global foresight, and shorter ones that improve local control but struggle with long-term tasks. To address this, a mixture of horizons (MoH) strategy is proposed, allowing for parallel processing of action chunks with varying horizons.

ActDistill has been introduced as a general action-guided self-derived distillation framework aimed at enhancing the efficiency of Vision-Language-Action (VLA) models. This innovative approach focuses on transferring action prediction capabilities from a well-trained VLA model to a lightweight version, addressing the computational overhead and inference latency that limit robotic manipulation applications.

The KV-Efficient VLA introduces a model-agnostic memory compression technique aimed at enhancing the efficiency of Vision-Language-Action (VLA) models by utilizing a recurrent gating module to selectively retain high-utility context during inference. This method addresses the computational challenges posed by traditional attention mechanisms and the extensive memory requirements for key-value pairs, particularly in long-horizon tasks.

The Evo-0 model has been introduced as a Vision-Language-Action (VLA) framework that enhances spatial understanding by integrating implicit 3D geometry features. This advancement addresses the limitations of existing Vision-Language Models (VLMs), which often lack precise spatial reasoning due to their reliance on 2D image-text pairs without 3D supervision.

AVA-VLA is a newly proposed framework aimed at enhancing Vision-Language-Action (VLA) models by integrating Active Visual Attention (AVA) to improve visual processing in dynamic decision-making contexts. This approach addresses the limitations of traditional VLA models that operate independently at each timestep, which can hinder effective contextual understanding in sequential tasks.

Recent advancements in Vision-Language-Action (VLA) models have led to the introduction of VLA-Fool, a study that investigates the adversarial robustness of these systems under both white-box and black-box conditions. This research highlights the vulnerabilities of VLAs, particularly in the context of cross-modal misalignment that can hinder decision-making processes.

VLA-Pruner has been introduced as a novel method for token pruning in Vision-Language-Action (VLA) models, addressing the inefficiencies of existing approaches that focus solely on semantic salience. This method aims to enhance real-time deployment of VLA models by retaining critical information necessary for action generation while discarding redundant visual tokens.