Indian traffic condition is very different from that of most western countries, the most glaring difference being the variety of vehicle types vying for road space at any given instance. The diversity arises from non-standardized physical size, speed/acceleration capacity and category (public transport versus private) of the vehicles. This leads to chaos in traffic flow resulting in unintentional non-adherence to rules and not necessarily a behavior trait. Usual traffic management policies suggest enforcement of strict lane discipline, though most city roads in India have only 1.5 to 2 lanes catering to the heterogeneous traffic. In our research, we focus on studying the driver response as a function of lane discipline conditions given the vehicle diversity. A road scenario similar to the typical Indian condition is simulated in Unity3D game engine platform and functionalities of the individual vehicle type are implemented as artificial intelligent (AI) bots. These bots interact with the player/first-person driver forcing her/him to react to the emergent situations. The scenarios are: a) no-lane discipline enforced and overtaking a slow-moving leader vehicle, b) lane change possible only at certain stretches of the road and the participant driver is positioned behind a slow-moving vehicle and c) when lanes are demarcated by solid structures and the driver is stuck behind a slow-vehicle. The factors of interest to this study are driver’s decision making and driving skills in each condition, the traffic disruption in each condition and the driver behavior – changing lanes without checking for clearance and self-report of anger/frustration. The scenarios are created in a virtual reality (VR) environment and the player controls the car in that simulated environment. Unity3D software was used to develop the application and Oculus/HTC Vive for VR. The player is provided with haptics (steering wheel, brakes) of a car so that he/she can physically feel that they are inside a car. Traffic sounds are added for natural effect. Experiments were done by participants with driving experience and also first time drivers.Time taken to reach the destination and the lane-change strategy was analysed for each condition. The time for no-lane demarcation condition was the least, while designated lane- change stretches required quick responses and speed-prediction skills, in the absence of which collisions were observed. The issues with a policy of lane-discipline without addressing the heterogenous traffic is highlighted. The findings from the preliminary understanding in real-life conditions and behavior in the virtual simulation led to a solution for road infrastructure model that is being discussed around the world and might be applicable for Indian city traffic.
Mayank Agrawal and Kavitha Vemuri
International Institute of Information Technology, Hyderabad
General, Urban Mobility