High Coverage Point-To-Point Transit: Hybrid evolutionary approach to local vehicle routing

High Coverage Point-to-Point Transit (HCPPT) is a new design of alternative transportation, which involves a sufficient number of deployed small vehicles in real-time response. This paper focuses on a hybrid evolutionary approach to improve the existing local vehicle routing algorithm of HCPPT. A hybrid Genetic Algorithm (GA) is proposed by utilizing an insertion heuristic method as a genetic operator. First, two genetic operation schemes, Random Feasible Position (RFP) and Best Feasible Position (BFP), are designed and tested on a simulation framework that we have developed recently. The simulation result shows the effectiveness of BFP in terms of system performance and computational efficiency. Next, we investigate a combined scheme that builds the initial BFP population based on an insertion heuristic. The second simulation reveals that the proposed initial population method provides considerably better solution qualities over the various problem constraints. The simulations in this study are performed with various demand levels based on SCAG (Southern California Association of Governments) transportation network and OCTA (Orange County Transportation Authority) trip demands.