Towards Evaluating the Benefits of Inter-vehicle Coordination

While vehicle automation has the potential to significantly improve safety and traffic efficiency, the full potential will only be realised when vehicles start exploiting wireless communication to cooperate with each other and coordinate their interactions in advance. Ensuring that vehicles coordinate safely while improving efficiency is, however, a very challenging problem as it depends on (i) the characteristics of individual vehicles (vehicle physics, sensors), (ii) unreliable wireless communication, and (iii) driving behaviour at a microscopic level, and their compounded effects at scale. The presence of non-communicative, non-automated vehicles must also be considered. Designing and evaluating coordination protocols requires a scalable simulation framework that is accurate both microscopically (to assess safety) and macroscopically (to evaluate efficiency). Standard car-following models, where position and velocity are dictated by local input stimuli, produce sometimes unrealistic behaviour when laterally changing position, and lack support for additional inputs. Furthermore, conventional environments used to model traffic flow are either too fine-grained to scale or too coarse to appropriately simulate control logic. This paper introduces RoundaSim consisting of (i) a traffic simulator using a novel approach of mixed discrete-continuous modes of time, and (ii) a framework for implementing car-following models that supports lane-changing and coordination protocols, with additional inputs from advanced sensors and wireless communications. We show how our framework can be used to implement and evaluate a car-following model with lane changes and validate that the traffic flow achieved approximates that of real-world highways. This allows our platform to be used as a baseline for evaluating the safety and efficiency of coordination protocols.