Composite robotic system for intelligent chemical experiment operations based on skill acquisition and knowledge graphs

Purpose The construction of automated laboratories will drive technological innovation in the field of radioactive and toxic volatile chemical synthesis. This study aims to propose a composite robotic intelligent control system for automated chemical experiments, replacing manual execution of experimental tasks. Design/methodology/approach A method for acquiring robotic experimental operation skills based on teleoperation is first proposed. By effectively extracting trajectory key points through the integration of geometric features and the robot’s kinematic characteristics, the impact of data noise on skill learning is reduced, thereby enhancing the robot’s ability to acquire operational skills. Meanwhile, a multilayer knowledge graph for robotic operation skills based on task retrieval is constructed, enabling fast search for experimental subtasks and operation skills. Finally, a relative pose visual-assisted positioning method based on ArUco markers is proposed, significantly improving the robot’s operational accuracy on chemical instruments. Findings Experimental results show that the method proposed in this paper can quickly acquire robotic experimental operation skills from teleoperation data, with an average trajectory point filtering rate of 91.42%. In addition, it enables rapid task and skill layer link searches using the multilayer knowledge graph, and, combined with the visual-assisted positioning method, effectively improves the composite robot’s operational accuracy on chemical instruments to 1.3 mm. Originality/value This research provides an effective method for intelligent control of composite robots aimed at chemical experiment, integrating teleoperation-based skill acquisition, rapid task and skill search using a multilayer knowledge graph and visual-assisted positioning, achieving precise operation of chemical instruments.