IISc's ARTPARK got permission to pilot a 5G-powered AI system on the Bengaluru-Hassan highway. The InterSafe project uses cameras and sensors to detect collision risks at junctions. "We have obtained permission to conduct the pilot on NH75," said Sheela Prabhakar. The system sends warnings to drivers through the mobile network.

InterSafe aims to detect collision risks where smaller roads meet the highway and send warnings to road users through the mobile network.

IISc's startup incubation and accelerator programme AI and Robotics Technology Park, popularly known as ARTPARK, has obtained permission to pilot a 5G-powered roadside artificial intelligence (AI) system on the Bengaluru-Hassan stretch of NH75.

The pilot, titled InterSafe, is expected to identify potential collisions at highway junctions, according to Sheela Prabhakar, who heads ARTPARK's Wired and Wireless Technologies Centre of Excellence.

Speaking to ET AI at the India Mobile Congress (IMC) 2026, she described InterSafe as a proof of concept being pursued with the National Highways Authority of India (NHAI).

"We have obtained permission to conduct the pilot on NH75, the Bengaluru-Hassan highway," she said.

Warnings at highway junctions Prabhakar said the team was working with IISc's traffic-safety specialists to address risks at unprotected junctions, where smaller roads meet the highway. People may walk onto the road, while cyclists, two-wheelers and cars may enter and turn in either direction, catching highway users unaware, she said.

"The objective of InterSafe is to see whether we can deploy roadside units with cameras, radar and other sensors to detect the possibility of a collision and give warnings to highway users," she explained.

In its published project description, ARTPARK said AI analyses live video and sensor readings to identify hazards, with warnings reaching vehicles and smartphone users over the cellular network. It calls the design an Infrastructure-to-Network-to-Person (I2N2P) model, linking roadside equipment, the network and the driver.

She added the system must issue selective alerts instead of repeated warnings not applicable to a driver as that could dissuade them from paying heed to the alarm.

"If drivers receive repeated alarms that are not relevant, there is a good possibility that they will ignore them," she said, adding why the system needs to issue selective alerts.

NHAI's 2023 traffic-management policy provides the broader context for such work, with provisions for AI-incident detection and future 5G communication as coverage increases, according to a road transport ministry statement.

The ministry said the standards include systems designed to recognise hazards such as animals and pedestrians on the highways, with alerts sent to patrol vehicles, message boards or nearby travellers.

Keeping equipment affordable Prabhakar said equipment costs are important because roadside units face vandalism and harsh environmental conditions, which require frequent repairs.

"With 5G, we can keep the roadside units lightweight and low-cost," she said, explaining that the network allowed the team to move computation away from equipment on the highway.

"All the AI processing can happen at what we call the edge, which is closer to the 5G infrastructure," she said.

The network's bandwidth and short transmission delays allow sensor data to reach nearby computing infrastructure, reducing the processing hardware and potentially lowering their costs and battery consumption, Prabhakar said.

She said startups, industry players and support from the wider ecosystem would be needed to take demonstrated technology to market or build products around it.

"We develop the technology, demonstrate it, and take it to a stage where startups or other industry players can take it forward," she added.