An AI traffic management tool developed by three IIT-Patna students has won Flipkart GRiD 2.0 after demonstrating the potential to reduce unnecessary deployments of traffic personnel and barricades by 86% during incident response.
The system, called Gridlock Oracle, was created for the innovation challenge’s theme of event-driven traffic congestion. It is designed to help traffic control rooms assess whether a reported incident is likely to escalate into a major disruption requiring road closures.
The winning team comprises Kashish Rai and Bhumi Garg, both students of mathematics and computing, and Pragya Mahajan of computer science. All three are in the third year of IIT-Patna’s BTech-MTech integrated dual-degree programme.
The national-level challenge was organised in collaboration with Bengaluru Traffic Police, which provided participants with a dataset of 8,057 real-world traffic incidents. The dataset allowed teams to develop and evaluate systems against actual incident patterns rather than relying only on simulated traffic scenarios.
Traffic authorities frequently have to make deployment decisions before the severity of an incident is clear. A reported crash, obstruction or other disruption may require traffic personnel, barricades and road closures, but an initial assessment may not indicate whether the situation will worsen. The uncertainty can lead authorities to position additional resources as a precaution, even when the incident does not ultimately require them.
Gridlock Oracle is intended to support that decision-making process by predicting which incidents are likely to require road closures. It also identifies suitable locations for positioning traffic management resources, according to details reported by the Times of India and a statement attributed to IIT-Patna registrar A K Thakur.
The solution was evaluated on previously withheld data, meaning the assessment used incident records that had not been used during development. The system demonstrated the potential to reduce unnecessary resource deployments by 86% in that evaluation, according to the report. The figure indicates the tool’s performance in the challenge assessment; the source does not establish that the same reduction has already been achieved in live traffic operations.
The project addresses a practical problem in urban traffic management: control rooms must balance rapid intervention against the cost and disruption of deploying personnel and barricades unnecessarily. Excessive deployment can tie up staff, occupy road space and divert response capacity from other incidents. At the same time, underestimating a developing disruption can delay road closures or traffic diversions.
The winning team’s system is therefore positioned as a decision-support tool rather than a replacement for traffic officials. Its proposed use is to help authorities determine the likely severity of an incident and plan the placement of resources more precisely. The report does not provide details on a live deployment, integration with a traffic control room or a timeline for operational adoption.
The result also highlights the role of urban datasets in developing tools for city administration. The Bengaluru Traffic Police dataset gave participants access to incident-level information linked to a specific traffic-management context. Any future operational use would depend on how such systems perform across different cities, incident types and data conditions, though the supplied report does not provide comparative results.
For now, Gridlock Oracle’s documented outcome is its first-place finish in Flipkart GRiD 2.0 and its performance on the withheld evaluation data. Further details on field trials, official adoption and the system’s effectiveness in live traffic conditions were not provided in the report.
