AI Can Predict Accident-Prone Areas, Save Lives

Hyderabad: A city-based research body has quickly identified accident-prone spots on highways so that remedial measures can be taken before things take a turn for the worse.

These accident-prone areas are known as black spots. A location is typically classified as a black spot if it witnesses frequent accidents and over 10 fatalities within three years.

However, advancements in artificial intelligence (AI) now eliminate the need to wait for three years to identify such areas.

“AI can predict black spots while they are still forming,” said Govind Krishnan, programme manager of iRASTE (Intelligence Solutions for Road Safety through Technology and Engineering) at the INAI, the Applied AI Research Centre at IIIT Hyderabad.

The iRASTE project is an initiative that leverages AI to address road topographical issues, provide real-time alerts to drivers to prevent accidents and enhance driving skills.

Launched by INAI in collaboration with Intel and the Telangana government, the project conducted a study in Telangana between April 2023 and March 2024. This study covered three key highways, from Hyderabad to Kodad (NH-65), Pullur and to Adilabad (NH-47).

As part of the initiative, Advanced Driver Assistance System (ADAS) devices and 10 driver monitoring system (DMS) units were installed on 200 buses operated by TSRTC. These devices provide real-time road data, playing a critical role in developing grey spot prediction models.

iRASTE has integrated four types of data to create statistical models for predicting grey spots: data covering 691 km of national highways, sourced from local and national road authorities; crash reports, including 5,606 FIRs and road crash records from 2022 to 2024, provided by the Telangana traffic police; ADAS alerts, collected quarterly since September 2022, and safety audits conducted at known black spots.

The study corridors were divided into 500-metre x 500-metre grids, and geographic information system (GIS) tools were used to calculate the number of ADAS alerts for each cell and identify road geometry parameters.

Regression models applied to this data pinpointed 20 consistent grey spots in each study corridor during the third and fourth quarters of 2023. The 80-90 per cent similarity in grey spots across these periods highlighted the persistent factors contributing to high crash severity, emphasising the importance of continuous monitoring and targeted safety measures.

“Combining different data sets, including live data from ADAS, has helped us identify several factors that contribute to grey spots,” explained Jonnada Prithvi, operations manager of iRASTE.

Through its modelling, the team identified contributory factors such as frequent traffic rule violations, damaged road infrastructure, and the lack of pedestrian safety measures, transverse bar markings, and crash barriers.

“We’ve found 20 grey spots on each route and these findings highlight the urgent need for targeted interventions,” he said.

Suggesting remedial measures for grey spots, iRASTE has submitted detailed project reports for 15 grey spots to the National Highway Authority of India (NHAI).

“While some grey spots can be rectified with simple solutions such as adding crash barriers, median studs, merging signboards and T-intersection warning signs, others require topographical changes. So far, the NHAI has invited tenders for the rectification measures of three grey spots and work is on the other spots,” informed Prithvi.

Additionally, iRASTE has trained 600 locals in active bleeding control (ABC) to act as first responders in case of accidents. “Our data shows that these first responders have saved 10 lives in just eight months,” Govind Krishnan added.

“This integrated preventive approach can predict black spots while they are still forming and save thousands of lives annually. The concept is scalable, and we are in discussions with the governments of Rajasthan and Jammu and Kashmir about extending the project,” he added.