Intelligent Driver Guidance Dashboard Framework to Prevent Road Accidents in Poor Visibility Conditions

Road accidents pose a considerable global public concern, particularly on highways and rural roads. Road accidents contribute to half of all traffic fatalities. In India, low visibility conditions accounted for 7.1% of the 2022 road accidents totalling 4,61,312 [1]. Road accidents are caused by various factors, including fog, rain, low light conditions, poor road maintenance, traffic rule violations, and driving under the influence. Among these factors, low-light conditions challenge even skilled drivers, as they struggle significantly to detect and avoid on-road obstacles while driving. Existing driver assistance systems have limitations under poor visibility conditions, often lacking accurate obstacle information. In this paper, an Intelligent driver guidance dashboard (IDGDB) is proposed to assist drivers in avoiding on-road obstacles such as four-wheelers, two-wheelers, pedestrians, traffic light poles, and trees under poor visibility conditions caused by low-light, and opposite headlight glare. The proposed dashboard uses Mask R-CNN (Mask Region-based Convolutional Neural Network) for instance image segmentation to detect, localize and classify the on-road obstacles. In addition to the image segmentation, an object-distance measurement algorithm (ODMA) is developed to calculate the distance between on-road obstacles and the Ego Vehicle (EGV). The segmented on-road obstacles and their distances are plotted on the driver guidance window (DGW) of the dashboard. The proposed IDGDB is tested in VIT Vellore campus roads in both daylight and low-light conditions. The experimental results are satisfactory as they are very close to the ground truth values. By providing real-time information, IDGDB empowers drivers to make informed decisions, contributing significantly to the reduction of road accidents and fostering safer road environments during poor visibility conditions.