From Documented Strengths to Force Limits: Material-Informed Robotic Insertion for Construction Assembly

Lin He; Yanyi Chen; Haofei Sun; Lingyao Li; Min Deng

cs.ROSep 18, 2026

From Documented Strengths to Force Limits: Material-Informed Robotic Insertion for Construction Assembly

Authors: Lin He, Yanyi Chen, Haofei Sun, Lingyao Li, Min Deng

Organizations: Department of Civil and Environmental Engineering, The University of Tennessee, Knoxville, TN 37996, USA · The University of Texas at Arlington, Arlington, TX 76019, USA · The University of Arizona, Tucson, AZ 85721, USA

Abstract

Insertion is a fundamental operation in robotic construction assembly, where variations in material properties and assembly conditions make it difficult to select contact forces that complete the task without exceeding the assembly's capacity. Although construction documents encode engineering knowledge about materials and their conditions, translating this knowledge into load limits for a specific assembly remains difficult. This paper presents SAGE (Source-grounded Assembly Gating and Execution), a system that converts documented material evidence into capacity estimates for robotic insertion. SAGE restricts a large language model (LLM) to extracting tensile and compressive strengths from retrieved passages and tables and records their sources. A response model then interpolates offline finite element (FE) solutions to convert these strengths and the assembly conditions into axial load capacity. For fits with positive clearance, the estimated capacity sets the policy's axial force limit; for interference fits, it is compared with measured support demand to determine admission. On the primary benchmark, SAGE reduces mean capacity error from 80.65% for direct LLM estimates based on the same evidence to 10.74%. Without refitting, the mean error remains 8.00% on 16 additional geometries. Under the assigned support release model, SAGE correctly classifies 59 of 62 scored simulation runs, with only conservative errors. In recorded xArm6 demonstrations, SAGE takes material documents as input and completes physical insertion in 9 of 13 trials. These results show that assigning document interpretation to the LLM and force calculation to an explicit mechanical model produces accurate capacity estimates and traceable insertion decisions.

Figures & tables

Fig. 1: Overview of SAGE. Documented strengths are converted to capacity under prescribed loading. The estimate sets the policy’s axial force input for positive clearance or is compared with support demand before an interference fit is dispatched. The workcell and force trace show the same recorded simulation of the prior insertion policy [ 6 ] ; these workcell visuals are illustrative. The photographs are recorded xArm6 hardware runs.

Fig. 2: Capacity estimation with known material strengths. (a) Response estimates versus refined FE capacities for four AAC grades at eight initial geometries; the diagonal denotes exact agreement. (b) Absolute relative errors of the response model and thick cylinder approximation for initial geometries g00–g07. (c) Errors for additional Sobol geometries c00–c15.

Item	Configuration or result
Fixed geometry	Bore radius $a=5$ mm, height $H=30$ mm, loaded span centered at $z=18$ mm, Poisson’s ratio $0.2$ .
Lamé verification	Radial resolutions: 8, 16, and 32 elements. Maximum stress error: 5.62% to 1.55%; displacement error: 0.0827% to 0.00521%.
Reference meshes	$32\times 192$ and $64\times 384$ . Maximum capacity change: 0.23% on the initial geometries and 2.19% on the additional geometries.
Response library	27 unit-pressure solutions over $t_{\mathrm{w}}\in\{1.5,3,6\}$ mm, $w\in\{6,12,18\}$ mm, and $\mu\in\{0.3,0.6,0.9\}$ .

TABLE I: Capacity evaluation configuration and reference checks.

Method	Valid $\uparrow$	Mean (%) $\downarrow$	Median (%) $\downarrow$	Max (%) $\downarrow$	Within 20% $\uparrow$
Primary estimators
Identity prior, direct LLM	80	79.95	58.78	2077.36	10
Identity prior + response	80	41.21	44.55	95.86	17
Linked evidence, direct LLM	80	80.65	41.12	1534.96	19
Linked evidence + calculator	80	90.47	41.13	2326.72	19
Linked evidence + FE tool	78	81.86	31.87	2345.71	33

TABLE II: Capacity estimation over 80 records from 40 material–geometry conditions. Bold marks the best result in the primary group, including ties.

Fig. 3: Language model diagnostics. (a) Capacity-error coverage; the dotted line marks 20%. (b) Complete strength recovery from four additional sources. (c) Recovery and abstention for direct tensile strength.

Fig. 4: Admission under the assigned support release model. (a) Original sleeve traces at an 11 N force limit. (b) Original sleeve support ladder; 7.0 N releases and 7.5 N holds at 3 and 11 N. (c) Outcomes for 40 stiff-sleeve cases. (d–e) Admission decisions at

B=15

N, reported as correct / conservative refusal / admitted release. Two LLM seeds share each simulation run; timeouts are reported separately.

Robotic Construction

Robotic construction develops robots, planning methods, and sensing systems for assembling structures and performing work on construction sites.

See all 53 papers

FORGE-plus: Force-Budgeted Recovery for Contact-Rich Assembly with a Frozen LLM Supervisor
Jul 23, 2026Kyupaeck Jeff Rah, Midum OhReinforcement Learning Language Model-Based Control
Independent Researcher
Generalizable Robotic Insertion with World Models
Sep 23, 2026Nicklas Hansen, Iretiayo Akinola, Yijie Guo +7Zero-Shot Robotic Manipulation World Models for Robotics
NVIDIA · University of California San Diego · Work completed during an internship at NVIDIA +1
InsertAnything: Generalizable Contact-Rich Precision Insertion from Simulation to Reality
Sep 21, 2026Zhenghua Ma, Xinpan Meng, Zeyu Liu +4RL for Robotics Contact-Rich Robotic Manipulation
School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049, China, and the State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China · PaXini AI Technology (Beijing) Co., Ltd., Beijing 100080, China

Jul 23, 2026J/K move · Enter open · S save

FORGE-plus: Force-Budgeted Recovery for Contact-Rich Assembly with a Frozen LLM Supervisor

Kyupaeck Jeff Rah, Midum Oh

Independent Researcher

Reinforcement Learning Language Model-Based Control Contact-Rich Robotic Manipulation Robot Failure Recovery Constrained RL Robotic Manipulation+1

Open paper AskPDF arXiv

Force-conditioned reinforcement learning (RL) enables tight-clearance assembly under a commanded force ceiling, but practical deployment requires determining an appropriate force limit for each object and recovering from insertion failures without exceeding it. We present a two-layer framework in which a frozen, text-only large language model (LLM) assigns a per-object force ceiling before execution and selects recovery maneuvers from a fixed action menu using compact textual force signatures. The LLM never controls force directly: a low-level controller enforces the force ceiling, the recovery policy cannot increase it, and the hidden breaking-force threshold is known only to the evaluator. We evaluate the framework on fragile bottle placement and 0.4 mm diametral-clearance gear insertion using two grippers (Robotiq 2F-140 and Franka Panda hand). A single policy passes 256/256 evaluation episodes on both fragile and robust objects without breakage, correctly predicts release timing, and completes a full table-pick-and-insert pipeline with a mean peak force of 5.4 N. Under injected in-grip slip, the force-signature recovery strategy resolves 40% and 64% of failures on the two grippers, whereas a press-harder baseline is either ineffective or causes frequent breakage. We also report negative results, including the failure of PPO to solve the task under strict force constraints and unsuccessful learned release strategies. All experiments are conducted in rigid-body simulation with hidden force-threshold breakage; no sim-to-real claim is made.

Abstract

Robotic Construction

Explore similar work

FORGE-plus: Force-Budgeted Recovery for Contact-Rich Assembly with a Frozen LLM Supervisor

Generalizable Robotic Insertion with World Models

InsertAnything: Generalizable Contact-Rich Precision Insertion from Simulation to Reality