EDGE ARTIFICIAL INTELLIGENCE BASED AUTOMATION FOR ULTRA-LOW-LATENCY CONTROL IN INDUSTRIAL ROBOTIC SYSTEMS

Abstract

This quantitative, cross-sectional, case-based study addressed the problem that cloud-connected industrial robotic systems often struggle to sustain ultra-low-latency closed-loop control because perception and decision workloads, network transit, and controller integration overhead introduce delay and jitter that can reduce motion precision and operational safety. The purpose was to quantify Edge AI based automation maturity and test whether it predicts perceived ultra-low-latency control performance (ULLCP) in industrial robotic cells using edge-to-cloud and enterprise OT/IT architectures. Data were collected from N = 162 practitioners (33.3% robotics or automation engineers, 28.4% operators or technicians, 19.8% maintenance, 18.5% OT/IT) across cloud and enterprise integrated robotic workcells. Key variables were Edge AI Automation (EA overall and four dimensions: local inference, real-time edge processing, controller integration readiness, reliability or failover readiness) and ULLCP (responsiveness, timing consistency, robustness), with task complexity and exposure level as controls. The analysis plan applied descriptive statistics, internal consistency reliability, Pearson correlations, and multiple regression. Findings showed high perceived EA maturity (M = 3.84, SD = 0.62) and high ULLCP (M = 3.77, SD = 0.58), and the scales were reliable (EA alpha = 0.91; ULLCP alpha = 0.88). EA correlated strongly with ULLCP (r = 0.61, p < 0.001). In regression, EA significantly predicted ULLCP (beta = 0.58, t = 9.42, p < 0.001) controlling for task complexity and exposure, explaining 46% of the variance (R2 = 0.46; F(4,157) = 32.94, p < 0.001). A dimension model explained 48% of variance (R2 = 0.48) and identified local inference (beta = 0.27, p = 0.001), controller integration (beta = 0.19, p = 0.012), and reliability or failover (beta = 0.22, p = 0.004) as the strongest contributors; real-time processing was positive but marginal (beta = 0.13, p = 0.058). These results imply that organizations seeking deterministic, low-latency robotics should prioritize near-device inference, stable controller interfaces, and resilient failover mechanisms alongside edge compute capacity, treating integration and reliability as first-class performance levers in Industry 4.0 modernization. Local inference (M = 3.92) rated highest, while integration (M = 3.73) and failover (M = 3.69) lagged; task complexity slightly reduced ULLCP (beta = -0.14) in practice.