Research Projects

Commercial bus drivers play a critical role in public road safety as they spend significant time behind the wheels. Driving in a complex traffic environment, exposure to long monotonous roads, inflexible work hours and prolonged driving may contribute to mental workload and driving stress, that may lead to higher accident risks and long-term health effects of long-haul bus drivers. Although a direct causal relationship between mental workload and accident involvement has not been well-established in the safety literature, workload has been linked to aggressive violations, risky driving attitudes, slower reaction times, negative long-term health effects of bus drivers and higher crash involvement rates. Majority of the existing studies have considered subjective assessments within controlled simulator environments to understand the possible relationship between drivers’ workload and accident involvement rates, however such an approach cannot capture workload fluctuations over time and its associated conflict risks in real-life traffic environments. To address this gap, this project attempts to develop an integrated proactive AI driven conflict-risk assessment framework considering drivers’ physiological state, driving style, surrounding traffic information and other external factors for predicting joint risk probabilities associated with long-haul bus drivers. Initially, an extensive questionnaire survey using the extended NASA-Total Load Index will be conducted to assess drivers’ mental workload, situational awareness and their driving performance under varying driving environments. Secondly, to capture fluctuations in mental workload and changes in driving behaviour over time, instrumented buses will be deployed in collaboration with State Transport Corporations across intercity routes. Each instrumented vehicle will be equipped with GPS receivers, IMU sensors, video cameras, wrist wearables and edge computing platforms to integrate naturalistic driving data from a minimum of 50 drivers. To support the research questions and objectives of the project, driver’s mental workload will be classified using physiological data such as electrocardiogram, electrodermal activity, gaze, blink duration, etc. based on an unsupervised clustering approach. These findings will be cross-validated against subjective assessments. The next step involves quantification of how mental workload, driving style and other factors contribute to conflict risks. Surrogate safety indicators will be used to compute conflict counts and severity levels. These metrics will serve as inputs to a copula-based modelling framework, enabling the estimation of joint risk probabilities under varying traffic and driver state scenarios. Finally, the applications of the copula model in road safety assessment will be discussed based on scenario analysis, and several recommendations for improvement in safety will be proposed. Sensitivity analyses will simulate diverse road geometries, traffic densities, and environmental conditions to evaluate conflict risks. Validation will be carried out in West Bengal using the same setup to assess model generalizability. Additionally, an expert opinion survey involving policymakers, fleet operators, and safety professionals will be conducted to prioritize strategies for improving safety. This project is expected to deliver the first high-resolution, multimodal naturalistic driving dataset for commercial buses in India, integrating physiological, kinematic and surrounding information. The outcomes will contribute to the development of cost-effective in-vehicle safety systems that enable real-time monitoring, physiological sensing and proactive risk alerting through V2V/V2I communication. The insights gained from this research will provide actionable guidance to national stakeholders, including MoRTH, NHAI, State Transport Undertakings on future policies, safe vehicle design, and intelligent system adoption in public transport fleets.