Real-Time Helmet Detection and Closed-Loop Enforcement System for Two Wheelers Automotive Research Association of India

Format:

Book

Conference/Event

Author/Creator:

Kandimalla, Om Mahalakshmi, author.

Contributor:

Karle, Ujjwala

Shah, Ravindra

Conference Name:

Symposium on International Automotive Technology (2026) (2026-01-28 : Pune, India)

Language:

English

Physical Description:

1 online resource cm

Place of Publication:

Warrendale, PA SAE International 2026

Summary:

Asian countries capture a significant share of global two-wheeler usage, with India consistently ranking among the top three countries. 2 wheelers are a significant portion of road traffic and contribute heavily to the national burden of road fatalities. Despite regulatory mandates, helmet non-compliance remains widespread due to limited enforcement reach and behavioural inertia. The current strategies for enforcement, such as traffic policing or external camera-based surveillance, are reactive, infrastructure-dependent, are ineffective at scale. To address these limitations, we propose system that will detect if the user is wearing the helmet. The system is designed and packaged to be integrated into the 2-wheeler directly and then execute functions in real-time for helmet noncompliance. The software algorithm is an AI-powered, vision-based system that leverages deep learning techniques for helmet detection. This model is enforced with a custombuilt dataset accommodating cultural and regional variations. Further model is trained and optimized so that it also perform accurately under conditions, including variable lighting, occlusions, and diverse headgear styles commonly seen in the Indian context.The overall system is further optimized for low-power, real-time inference suitable for embedded platforms on two-wheelers. Once the helmet is not detected, the system generates a two-stage response: an audible alert warns the rider, and if non-compliance persists, the vehicle can trigger a controlled deceleration mode through a closedloop actuation strategy, bringing it to a safe stop. The evaluation results indicate a detection accuracy of 97% under varied real-world conditions, establishing the feasibility of intelligent, vehicle-integrated enforcement for two-wheelers in the Indian context

Notes:

Vendor supplied data

Publisher Number:

2026-26-0650

Access Restriction:

Restricted for use by site license