An Autonomous Modular Public Transit service

In this work, we present a proof-of-concept investigation of Autonomous Modular Public Transit (AMPT) at a network scale and compare it with the traditional fixed-route, fixed-vehicle size transit service in terms of total cost, which consists of both agency's capital and operational cost (including energy cost) and passenger time cost. We formulate and solve stylized design models for AMPT on a grid network in a range of demand density scenarios with both homogenous and heterogeneous distributions. The AMPT models explicitly account for pod joining and disjoining (and therefore en-route transfers of passengers) and potential energy savings due to pod train formation (pod platooning), which represent major departures from the traditional transit models in the literature. Numerical results find that AMPT, if designed properly, may save the total cost compared to traditional transit systems thanks to demand responsive pod train capacity, particularly in the low demand scenarios. The cost savings of AMPT are largely attributed to passenger time saving by en-route transfer; the agency cost of AMPT has a mixed picture. The load factor of AMPT generally improves over the traditional transit service. We also show how key parameter values may affect the AMPT costs through sensitivity analysis.