AgentChemist: A Multi-Agent Experimental Robotic Platform Integrating Chemical Perception and Precise Control
Authors: Xiangyi Wei, Fei Wang, Haotian Zhang, Xin An, Haitian Zhu, Lianrui Hu, Yang Li, Changbo Wang, +1 more
Organizations: School of Computer Science and Technology, East China Normal University, Shanghai, China. · School of Chemistry and Molecular Engineering, East China Normal University, Shanghai, China. · School of Data Science and Engineering, East China Normal University, Shanghai, China. · Shanghai Frontiers Science Center of Molecule Intelligent Syntheses, Shanghai, China.
Chemical laboratory automation has long been constrained by rigid workflows and poor adaptability to the long-tail distribution of experimental tasks. While most automated platforms perform well on a narrow set of standardized procedures, real laboratories involve diverse, infrequent, and evolving operations that fall outside predefined protocols. This mismatch prevents existing systems from generalizing to novel reaction conditions, uncommon instrument configurations, and unexpected procedural variations. We present a multi-agent robotic platform designed to address this long-tail challenge through collaborative task decomposition, dynamic scheduling, and adaptive control. The system integrates chemical perception for real-time reaction monitoring with feedback-driven execution, enabling it to adjust actions based on evolving experimental states rather than fixed scripts. Validation via acid-base titration demonstrates autonomous progress tracking, adaptive dispensing control, and reliable end-to-end experiment execution. By improving generalization across diverse laboratory scenarios, this platform provides a practical pathway toward intelligent, flexible, and scalable laboratory automation.
Figures & tables
Figure 1: The motivation for a multi-agent experimental robotic platform. Traditional rigid automation (left) is characterized by fixed protocols, blind execution, and fragile adaptation. In contrast, AgentChemist introduces dynamic workflows through multi-agent task decomposition, adaptive perception with real-time monitoring, and robust, instrument-integrated resilience to address long-tail challenges in the laboratory.
Figure 2: AgentChemist Workflow: AgentChemist receives user instructions, plans experimental protocols, drives robots to perform specific experimental operations, records results in real time, and finally returns the experimental report to the user.
Audio-based estimates Q^audio with confidence γa , statistical estimates Q^stat with confidence γs , fused high-confidence values Q^fused
D , high-confidence recorder
Table 1: Structured Knowledge Extracted for Experimental Report Generation
Figure 4: State Machine of the Statistics-based Chemical Quantity Logger
Figure 5: The Process of AgentChemist Driving the Evolution of FSM S in Acid-Base Titration Experiments
Figure 6: Acid-Base Titration Experiments. a Titration curve of maleic acid (concentration: 0.1M , initial volume: 25mL , initial pH=1.51 ). b-c Enlarged view of the transition region in the maleic acid titration curve. d-e First-derivative and second-derivative plots of the maleic acid titration curve. f Key conclusions from the Agent data analysis report for the maleic acid titration experiment. g-h Titration curves of hydrochloric acid and acetic acid (both at 0.1M concentration, initial volume: 25mL ). i Standard deviation plot comparing the hydrochloric acid titration curve with theoretical values (comparison of three parallel experiments and theoretical simulation).
Figure 7: Chemosensor with Color Perception. The color transition sequence (from left to right) during the complexometric titration of Ca2+ concentration using a 0.02M EDTA standard solution against a 0.01MCaCl2 solution (initial volume 5mL ) with a calcium indicator was as follows: the solution initially exhibits a magenta color due to the calcium–indicator complex, and finally turns sapphire at the endpoint upon the release of the free indicator.
No.
Weighing Error (g)
Time (s)
Success
1
0.08
45.2
Yes
2
0.11
47.8
Yes
3
0.05
43.5
Yes
4
0.15
52.1
Yes
5
0.09
46.3
Yes
6
0.07
44.9
Yes
Table 2: Experimental Results for Solid Weighing and Dissolution Task
Environmental Conditions
Success Rate
Average Weighing Error (g)
Standard Layout
98%
0.08
Disturbed Layout (Multiple Sizes of Beakers)
94%
0.11
Table 3: Generalization Performance of the Solution Preparation Task in Different Environments
Configuration
Task Completion Rate
pH Deviation
Anomaly Detection Accuracy
Control Group
96%
0.11
94%
Experiment Group A
92%
0.15
89%
Experiment Group B
78%
0.26
73%
Table 4: System Generalization Performance under Different Model Configurations
Figure 8: Configuration of the AgentChemist robotic experimental platform. The system features a Tracer 2.0 mobile base with all-terrain design and independent suspension for precise navigation. Two 7-DoF robotic arms (Piper Standard) are mounted on a rigid frame, offering 1.5 kg payload capacity and ±0.1 mm repeatability for precise laboratory manipulation. The perception system includes three Orbbec Dabai DC1 depth cameras (1920×1080, 30 fps) mounted on the wrists and front, plus two contact microphones embedded in the grippers to capture acoustic signatures during experimental operations.
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.
Wet-lab experimentation serves as the gold standard for hypothesis verification in scientific discovery; yet it is inherently labor-intensive, costly, and safety-critical. Embodied agents hold the promise of automating these tedious workflows, but their development is hindered by the scarcity of real-world training data. While simulation offers a scalable alternative for producing demonstrations, current methods primarily target relatively short-horizon tasks with loosely structured interactions, failing to meet the strict procedural constraints and fine-grained manipulation demands of chemical experiments. To bridge this gap, we introduce RoboChemGym, a framework that autonomously generates high-fidelity manipulation demonstrations aligned with real-world experiment protocols, featuring a self-improving task synthesis mechanism to iteratively refine task execution and scene configurations, enabling the reliable generation of expert trajectories for complex, multi-object protocols exceeding 10 interaction steps. Furthermore, we introduce a hierarchical benchmark that systematically assesses performance across varying granularities, spanning from atomic operations to full-cycle experimental workflows. RoboChemGym sets a scalable paradigm for the automated data synthesis and capability evaluation of embodied agents in intricate chemical tasks, serving as a critical stepping stone toward fully intelligent laboratories.
Chenxi Li, Haiyuan Wan, Rui Li +10
Shanghai Artificial Intelligence Laboratory · Zhejiang University · Tsinghua University +7
Automating science laboratories enables faster, safer, more accurate, and more reproducible execution of protocols, accelerating the discovery and testing of new materials, drugs, and more. However, setting up and running autonomous labs requires coordinating numerous instruments and robots, forcing scientists to write code, manage configuration files, and navigate complex software infrastructure. We present an AI agent architecture that integrates large language models with laboratory orchestration, enabling scientists to interactively create and monitor automated lab protocols using natural language. Integrated into the Experiment Orchestration System (EOS), the AI agent operates under an agentic loop with automated validation and error correction, and supports the complete experimental lifecycle: creating protocols, running and monitoring both protocols and closed-loop optimization campaigns, and analyzing results. A visual graph editor renders protocols as interactive node-based diagrams synchronized with the AI agent's protocol representation, enabling seamless alternation between AI-assisted and manual protocol construction. Evaluated on three simulated automated labs spanning chemistry, biology, and materials science, the AI agent achieves a 97% first-attempt protocol generation success rate and an order of magnitude reduction in required interface actions.
Angelos Angelopoulos, James F. Cahoon, Ron Alterovitz
Department of Computer Science, University of North Carolina at Chapel Hill, NC 27599, USA · Department of Chemistry, University of North Carolina at Chapel Hill, NC 27599, USA