CARE: Multi-Task Pretraining for Latent Continuous Action Representation in Robot Control
This addresses scalability and generalization issues in robot control for robotics applications, representing an incremental improvement over existing methods.
The paper tackles the challenge of Vision-Language-Action models' dependence on action supervision in robot control by introducing CARE, a framework that uses only video-text pairs for pretraining to learn continuous latent action representations, achieving superior success rates and interpretability in simulation tasks.
Recent advances in Vision-Language-Action (VLA) models have shown promise for robot control, but their dependence on action supervision limits scalability and generalization. To address this challenge, we introduce CARE, a novel framework designed to train VLA models for robotic task execution. Unlike existing methods that depend on action annotations during pretraining, CARE eliminates the need for explicit action labels by leveraging only video-text pairs. These weakly aligned data sources enable the model to learn continuous latent action representations through a newly designed multi-task pretraining objective. During fine-tuning, a small set of labeled data is used to train the action head for control. Experimental results across various simulation tasks demonstrate CARE's superior success rate, semantic interpretability, and ability to avoid shortcut learning. These results underscore CARE's scalability, interpretability, and effectiveness in robotic control with weak supervision.