Non-deterministic journey planning in multi-modal transportation networks

Abstract

Multi-modal journey planning, which allows multiple modes of transport to be used within a single trip, is becoming increasingly popular, due to a vast practical interest and the increasing availability of data. In real-life situations, transport networks often involve uncertainty, and yet, most approaches assume a deterministic environment, making plans more prone to failures such as significant delays in the arrival or waiting for a long time at stations. In this paper, we tackle the multi-criteria stochastic journey planning problem in multi-modal transportation networks. We consider the problem as a probabilistic conditional planning problem, and we use Markov Decision Processes to model the problem. Journey plans are optimised simultaneously against five criteria, namely: travel time, journey convenience, monetary cost, CO2, and personal energy expenditure (PEE). We develop an NSGA-III-based solver as a baseline search method for producing optimal policies for travelling from a given origin to a given destination. Our empirical evaluation uses Melbourne transportation network using probabilistic density functions for estimated departure/arrival time of the trips. Numerical results demonstrate the effectiveness of the proposed method for practical purposes and provide strong evidence in favour of contingency planning for journey planning problem.