Abstract
Drivers are essential to any industry offering transportation and logistics services. Ensuring their well-being ensures smooth business and reduces stress factors that cause driver burnout and increase fatigue and stress. Burnout is usually responsible for accidents, absenteeism, and other similar problems, that are best prevented. Therefore, understanding these stress factors and determining ways to overcome them would benefit any related industry. In the proposed approach, we leverage the benefits of natural language processing and the availability of numerous studies on driver burnout, fatigue, and stress to determine the various stress factors and to understand how to address those using a vehicle routing problem. First, topic modeling, a popular natural language processing technique, is used to extract the different topics of discussion around driver burnout, stress, and fatigue. Next, the extracted cases are qualitatively analyzed to ascertain the stress factors that can be controlled through a routing problem and how to do so. Since uncertainty is prevalent in real life, pairwise travel times are assumed to follow different functional forms. Finally, an integrated routing model is developed, and a hybrid genetic algorithm is coded to solve the model. The use of various sources and types of data, viz., structured data for routing and unstructured data for topic modeling, to obtain solutions for routing and driver well-being, simultaneously, is a notable contribution of the proposed approach. Also, the integrated use of qualitative and quantitative research methods and the combination of text analytics and combinatorial optimization to model social problems is a relatively new concept in the vehicle routing literature. Experimental studies on existing datasets provide solutions that illustrate the advantages of the approach. Insights are also provided to assist managers in decision-making under similar scenarios.