The rise in demand for taxi services such as Ola, Uber, Grab, Lyft, and Cabify, provides an economical and significant market in travel and tourism. Taxi service companies offer high incentives to taxi drivers for completing more-frequent trips. Therefore, drivers work long hours to increase their perks, but performing continuous journeys increases stress, which is an influential factor of distraction and leads to dangerous driving behavior and accidents. Motivated by the severity of increasing accidents and stressed driving, we propose a stress-aware recommendation system (SARS) that suggests the driver accept, reject, or compromise the upcoming trip on behalf of the current stress level. Most of the existing approaches depend on physiological sensors for stress detection and pay less attention to personalizing the stress according to one's capability. In this paper, we divided the proposed methodology into two phases. The first phase involves personalizing drivers' stress and behavior using a novel multi-task learning-based MTL-ConvLSTM. The second phase deals with finding a relationship between stress and behavior that helps predict the driving behavior from the stress level. The result shows that the proposed system can predict the stress level with an accuracy of 92 % and the driving behavior with 95 % accuracy, thereby making the methodology reliable.
Stress-Aware Recommendation for Safe Driving using MTL-ConvLSTM
2022-10-08
770446 byte
Conference paper
Electronic Resource
English
Spatiotemporal Bus Arrival Prediction Using ConvLSTM and CTGANs-augmented Data
Springer Verlag | 2025
|Spatiotemporal Bus Arrival Prediction Using ConvLSTM and CTGANs-augmented Data
Springer Verlag | 2025
|