Enhancing Bicycle Trajectory Planning in Urban Environments through Complex Network Optimization

This paper introduces a novel methodology for devising bike trajectory plans. In this approach, the entire trip is associated with a complex network or graph, where the nodes and edges correspond to potential trajectory segments. Utilizing the information collected from a bike-sharing system, the cost attributed to each segment is determined using the multinomial logistic regression (MLR) technique by analyzing its usage frequency and other user preferences. Consequently, segments with high usage and strong preferences entail lower costs, whereas those with limited use and weaker predilection assume higher costs. After assigning costs to all segments within the network, the subsequent step involves generating the trajectory with the lowest accumulated cost. Unlike other approaches, our method considers realistic conditions and user preferences without inconsistencies imposed by the techniques based on surveys. To validate the performance of the proposed method, a set of extensive experiments and case studies were conducted considering the urban model from downtown Guadalajara, Mexico. As a result, this approach improves the efficiency of the bike trajectory planning system by providing shorter and safer routes for both cyclists and motor vehicle drivers.