Long-Term Memory for VLA-based Agents in Open-World Task Execution

2026-04-16

AI Summary

This work addresses the limitations of existing vision-language-action (VLA) models in automating complex chemical experiments, particularly their inability to perform long-horizon reasoning and accumulate experiential knowledge for efficient multi-stage task execution. To overcome these challenges, the authors propose ChemBot, a novel framework featuring a dual-layer long-term memory architecture and a progress-aware Skill-VLA model. ChemBot integrates a Model Context Protocol (MCP) server to coordinate sub-agents and tool invocations, and employs a future-state-driven asynchronous reasoning mechanism to enable experience consolidation and continuous decision-making. Experimental evaluations on a collaborative robotic platform demonstrate that ChemBot significantly enhances operational safety, precision, and success rates in long-duration chemical experiments, outperforming current VLA baselines across all metrics.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Language and Vision
• Natural Language Processing: Safety and Robustness

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Vision-Language-Action (VLA) models have demonstrated significant potential for embodied decision-making; however, their application in complex chemical laboratory automation remains restricted by limited long-horizon reasoning and the absence of persistent experience accumulation. Existing frameworks typically treat planning and execution as decoupled processes, often failing to consolidate successful strategies, which results in inefficient trial-and-error in multi-stage protocols. In this paper, we propose ChemBot, a dual-layer, closed-loop framework that integrates an autonomous AI agent with a progress-aware VLA model (Skill-VLA) for hierarchical task decomposition and execution. ChemBot utilizes a dual-layer memory architecture to consolidate successful trajectories into retrievable assets, while a Model Context Protocol (MCP) server facilitates efficient sub-agent and tool orchestration. To address the inherent limitations of VLA models, we further implement a future-state-based asynchronous inference mechanism to mitigate trajectory discontinuities. Extensive experiments on collaborative robots demonstrate that ChemBot achieves superior operational safety, precision, and task success rates compared to existing VLA baselines in complex, long-horizon chemical experimentation.

Problem

• long-term memory
• VLA-based agents
• open-world task execution
• chemical laboratory automation
• long-horizon reasoning

Innovation

• long-term memory
• Vision-Language-Action (VLA)
• hierarchical task decomposition
• asynchronous inference
• embodied AI agent