.png){kind=logo}
Newsletter N°89 - May 2026
🏗️ Construction: SoftBank deploys private 5G and edge AI to prevent autonomous vehicle collisions in factories
SoftBank has detailed the results of a proof-of-concept project aimed at improving safety around autonomous industrial vehicles through private 5G, edge AI and real-time video analytics. Conducted with Sumitomo Electric and TechnoPro, the initiative focuses on preventing collisions between workers and autonomous transport vehicles operating in factory environments.
The project addresses a growing challenge linked to industrial automation. As factories increasingly deploy autonomous forklifts and transport systems to improve productivity and address labor shortages, safety remains a key concern. Conventional systems typically rely on onboard sensors and cameras, but these can struggle in crowded industrial environments filled with containers, shelves and narrow corridors where blind spots limit visibility.
To address this issue, the proof of concept introduces a distributed safety model based on fixed AI cameras installed throughout the factory. Positioned near blind corners and obstructed areas, these cameras continuously monitor activity and transmit video streams through industrial 5G terminals to a nearby MEC (Multi-access Edge Computing) server equipped with GPU processing.
The AI camera used for real-time collision detection.
The edge AI system analyses video locally and predicts collision risks between workers and vehicles in real time. When danger is detected, warning signals are immediately triggered to support avoidance actions.
A key outcome of the trial was the ability to perform collision detection and risk assessment with latency below 150 milliseconds, or approximately 0.15 seconds. According to SoftBank, this significantly improves reaction capability in factory environments where autonomous vehicles move continuously and operate in close proximity to workers.
The architecture also aims to improve deployment efficiency. By shifting AI processing and environmental monitoring from individual vehicles toward shared network infrastructure, operators may reduce reliance on expensive onboard sensors and minimize vehicle modifications. Network slicing further ensured that safety-related traffic received communication priority, maintaining stable and predictable performance even under congested conditions.
Private 5G and edge AI safety architecture.
The three companies contributed complementary capabilities to the system. Sumitomo Electric provided industrial 5G terminals, TechnoPro developed the collision-avoidance AI, and SoftBank supplied the Private 5G and MEC infrastructure supporting real-time processing.
Although tested in factory logistics environments, SoftBank positions the technology as a broader safety platform potentially applicable to other mobility scenarios, including road intersections and urban transport. Strategically, the initiative illustrates how telecom operators are increasingly moving beyond connectivity to provide integrated AI and industrial infrastructure services, positioning 5G as a real-time operational layer for automation and mobility safety.