cs.AIAug 13, 2026

Retrieval-grounded robot program generation and simulation-based correction via Model Context Protocol

Authors: Zhichao Zhou, Siyuan Chen, Omkar Salunkhe, Ebru Turanoglu Bekar, Johan Stahre, Anders Skoogh

Organizations: Department of Mechanical Engineering, Chalmers University of Technology, Hörsalsvägen 7a, Gothenburg, SE-412 96, Sweden

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Flexible manufacturing requires industrial robots to be reprogrammed rapidly as product variants change. This paper presents a language-model-based workflow that generates, validates, and iteratively corrects ABB RAPID robot programs from natural language task descriptions. A dual-stream retrieval-augmented generation (RAG) pipeline grounds code generation in verified technical documentation and production templates, reducing domain-specific errors produced by ungrounded language models. A custom Model Context Protocol (MCP) server connects the language-model client directly to ABB RobotStudio for automated code upload, simulation execution, and diagnostic feedback. The evaluation combines a 30-query retrieval benchmark, scoped code-generation checks, and RobotStudio case studies in a simulated pickand- place manufacturing cell. The simulation loop exposes execution failures that static and semantic checks alone cannot catch, including suction release-height errors, unreachable placement targets, and configuration-dependent recovery motions. The results show how RAG and MCP can connect grounded code generation with executable feedback from industrial robot simulation software, while reducing but not eliminating expert setup and final supervision.

Figures & tables

Figure 1: Overall retrieval-and-planning and validation-and-execution workflow.
Evidence itemUnitRole in the paper
Retrieval benchmark30 queriesDirect comparison of retrieval configurations using the same query set.
Pure LLM baseline5 modulesSmall diagnostic baseline for RAPID hallucination modes; not a paired statistical baseline for all generated modules.
RAG generation log21 modulesBroader check of structural validity and functional coverage after retrieval grounding.
Semantic review subset10 modulesSubset with logged full-pipeline review outcomes before simulation.
Release-height case1 cell taskMechanism evidence that simulation detects a suction release failure missed by text checks.
Pyramid variants2 tasksCase evidence for product-variant reconfiguration using the same RobotStudio cell.
Table 1: Evaluation scope and intended comparisons.
ConfigurationPrecisionRecallMRR
Naive RAG (documentation only)0.520.410.58
Dual retrieval with code templates0.680.570.72
Table 2: Retrieval performance averaged over 30 test queries.
Figure 2: Release-height failure exposed through RobotStudio simulation. The same +40 mm offset works for the 100 mm orange box but leaves the TCP inside the 200 mm green box. The corrected release offset is +140 mm.
Figure 3: Product-variant reconfiguration in the RobotStudio cell. The orange variant succeeded on the first attempt after scene introspection; the green variant required additional recovery because the changed pallet and box geometry exposed reach and joint-configuration constraints.
Figure 4: Joint-configuration recovery in the green-box variant. The successful correction used MoveAbsJ to reset the robot to a calibration pose and MoveJ to return to the named home pose without forcing the fifth axis back near its stored high-angle home pose.
More in this topic

Code Generation for Robotics

See all 20 papers

Explore similar work

CardsList

  1. RAPID: Robot Agentic Programming from Demonstrations

    Sep 24, 2026Yuyao Liu, Jiayuan Mao, David Hsu +2Program SynthesisContact-Rich Robotic Manipulation

    Massachusetts Institute of Technology · National University of Singapore · University of Pennsylvania +1

  2. Domain-Adaptable Reinforcement Learning for Code Generation with Dense Rewards

    May 20, 2026Erfan Aghadavoodi Jolfaei, Daniel Maninger, Abhinav Anand +2RL for Code GenerationReinforcement Learning

    Technische Universität Darmstadt, Darmstadt, Germany · Hessian Center for Artificial Intelligence (hessian.AI), Darmstadt, Germany · National Research Center for Applied Cybersecurity ATHENE, Darmstadt, Germany.

  3. Bridging Semantics and Physics with Constrained LLMs for Safe and Trustworthy Robotic Manipulation

    Aug 29, 2026Wenhao Hong, Lan Wei, Dandan ZhangSafe Robot ManipulationLanguage-Conditioned Robot Manipulation

    Imperial College London, London, UK