AI-Driven Collaborative Assembly Line Inspection: System Integration and Deployment Challenges

Asya Ünal; Amr Okasha; Ege Çırakman; Perin Ünal

cs.ROSep 27, 2026

AI-Driven Collaborative Assembly Line Inspection: System Integration and Deployment Challenges

Authors: Asya Ünal, Amr Okasha, Ege Çırakman, Perin Ünal

Organizations: Department of Research and Development, TEKNOPAR, Ankara, Türkiye

Abstract

Manual visual inspection on assembly lines is a persistent manufacturing bottleneck: operator fatigue over extended shifts lowers defect-detection rates. This paper presents the design, integration, and field deployment of an AI-assisted collaborative inspection cell at the Silverline kitchen-appliance factory, developed within the AI-PRISM project. The cell couples a Universal Robots UR 10e cobot carrying a machine-vision defect-detection pipeline with a Comau Racer-5 cobot for functional tests, coordinated through ROS 2 Humble on an Ubuntu 22.04 LTS server. Multi-modal data (Basler camera imagery, TIA microphone acoustics, and SPS electrical-safety measurements) are logged locally and visualised in real time with Grafana. We report the practical deployment challenges (close-proximity safety, AI robustness under glare and reflections, ROS 2 namespace collisions across two cobots, and operating-system and dependency issues) together with the engineering solutions adopted, and structure the integration through a four-level Human-Robot Interaction analysis. The deployed cell cuts per-unit quality-check time from 82 s to 61 s (about 25%), raises final-control resource efficiency from 0.75 to 0.88, reduces operator visual-inspection viewing time by 82%, and significantly lowers operator mental demand (p = 0.005, NASA-TLX).

Figures & tables

Figure 1: Complete system architecture of the Silverline inspection cell. Blue lines: Ethernet; red: ROS 2 communication; orange: power supply; purple: USB; green: Profinet/Profisafe; grey: digital I/O. The functional-safety subsystem (bottom) comprises a Siemens S7-1200 PLC, emergency-stop buttons, and an HMI panel. AI modules for defect detection and object tracking are hosted on the central server.

Figure 4: Still frame of the tracking AI module showing several white appliances tracked on the conveyor belt.

Component	Specification
Inspection cobot	Universal Robots UR 10e collaborative manipulator
Functional-test cobot	Comau Racer-5 collaborative manipulator, TCP/IP driver
End-effector camera	Basler ace industrial camera, USB 3.0, with anti-glare polarising lens and annular LED ring
Cell-overview camera	Dahua IP security camera
Acoustic sensor	TIA microphone, interfaced via signal conditioner and a LabJack DAQ
Electrical safety tester	SPS KT1885 electronic safety tester, serial interface

Table 1: Principal hardware components of the Silverline inspection cell

Indicator	Baseline	Achieved
Hood detection & tracking accuracy	—	$\approx$ 100%
2D defect-detection accuracy	—	$\approx$ 85%
Per-frame inference latency	—	$\approx$ 0.25 s
End-to-end perception latency	—	$\approx$ 0.5 s
Per-unit QC time (U4_KPI5)	82 s	61 s ( $-25$ %)
Resource efficiency (U4_KPI4)	0.75 (3 ops)	0.88 (1 op, $+17$ %)

Table 2: Key performance indicators of the Silverline inspection cell

Explore similar work

A Vision Based Framework Integrating Attention and Action Cues for Interpretable Cognitive Workload Assessment in Human-Robot Collaborative Assembly
Sep 14, 2026Junyan Xiong, Naiyi Feng, Xingke Xia +3Human-Robot Collaboration Assembly
Smart Manufacturing Thrust, System Hub, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China · College of Future Technology, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China
Harness Robotic OS: A Unified Embodied-Agent Runtime for Closed-Loop Quadruped Inspection
Sep 11, 2026Yaoyuan Yan, Zhiyou Heng, Haoxiang Jie +3Robot Operating System Inspection
AI Lab, Country Garden Services · Omni AI · East China Normal University
A Confidence-Aware Multimodal Fusion Framework for Industrial Human-Robot Collaboration
Sep 9, 2026Xinyu Liu, Qiqi Dong, Boya Jia +2Human-Robot Collaboration Multimodal Fusion

Sep 14, 2026, cs.ROJ/K move · Enter open · S save

A Vision Based Framework Integrating Attention and Action Cues for Interpretable Cognitive Workload Assessment in Human-Robot Collaborative Assembly

Junyan Xiong, Naiyi Feng, Xingke Xia, Qihang Fan+2

Smart Manufacturing Thrust, System Hub, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China · College of Future Technology, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China

Human-Robot Collaboration Assembly Robot Systems Cognitive Load Physiological Data

Open paper AskPDF arXiv

The introduction of human-robot collaboration (HRC) in industrial assembly operations is revolutionizing the manufacturing landscape. In this evolving environment, operators are required to seamlessly coordinate their manual tasks with real-time task information and robotic behaviors. These demands fluctuate during operation, yet conventional workload assessments depend on body-worn physiological sensors that complicate practical deployment. Here, we present a vision-based attention--action framework for continuous and interpretable workload-related assessment in HRC assembly. The framework combines RGB-D observations with robot states and calibrated task-related areas to construct a temporally confirmed representation of operator behavior. This representation identifies where task demand is concentrated and explains how it develops when attention and action diverge, the task context changes, or the operator hesitates. We evaluated the framework in a three-level collaborative gearbox assembly experiment with ten participants, using subjective ratings and synchronized physiological signals as independent references. Raw NASA-TLX ratings confirmed increasing perceived workload across conditions, with significant effects on overall workload and its mental and temporal dimensions. The vision-derived HRC-CWL output was significantly associated with ECG-derived features in seven of nine participants with complete correlation data. Synchronized interaction episodes further showed temporal correspondence between detected hesitation and physiological activity. Real-time deployment demonstrated that the framework can operate without requiring operators to wear additional sensors. These findings support HRC-CWL as an interpretable behavioral proxy for cognitive ergonomics analysis and adaptive robot assistance, rather than a direct psychophysiological measure of workload.

Abstract

Explore similar work

A Vision Based Framework Integrating Attention and Action Cues for Interpretable Cognitive Workload Assessment in Human-Robot Collaborative Assembly

Harness Robotic OS: A Unified Embodied-Agent Runtime for Closed-Loop Quadruped Inspection

A Confidence-Aware Multimodal Fusion Framework for Industrial Human-Robot Collaboration