Trinity College Dublin

Skip to main content.

Top Level TCD Links

Research

Prof. Cahill's research addresses many aspects of distributed systems, in particular, middleware and programming models for mobile, ubiquitous and autonomic computing with application to optimization of urban resource usage and service delivery in order to improve the quality of life and sustainability of cities. In the past, he worked extensively on middleware and programming models for distributed object computing. His current work is addressing support for connected and autonomous vehicles with the objective of optimizing journey time predictability in both highway and urban settings.

Research Supervision

Prof. Cahill supervises graduate students in the general area of support for connected and autonomous vehicles. Topics of current interest include:

  • Coordination Models for Connected and Autonomous Vehicles with the objective of optimizing journey time predictability in both highway and urban settings and specifically on coordination architectures for road vehicle automation
  • Journey Time Predictability for Connected and Autonomous Vehicles with the objective of optimizing journey time predictability in both highway and urban settings and specifically on the use of self-organizing algorithms for traffic system optimization.

Interested students should have a B.Sc. and/or M.Sc. in Computer Science, Computer Engineering or a closely-related discipline as well as exceptional C++, C#, and/or Java software development skills. Experience in distributed systems and/or machine learning is desirable as are strong mathematical skills.

Enquiries may be made to vinny.cahill@tcd.ie quoting CAV Studentship in the subject line.

Current Projects

  • Coordination Architectures for Road Vehicle Automation

    Large-scale deployment of (semi-)autonomous vehicles (AVs) is inevitable. However, the benefits of this deployment for traffic management in a world in which AVs and other vehicles will necessarily coexist remain unclear. Reduced congestion, greater energy efficiency, and improved resilience of the traffic system to unexpected events are expected. In this context, the hypothesis motivating this work is that full advantage of large-scale deployment of AVs for traffic management will only be achieved if AVs are designed to coordinate their behaviours with each other and with other vehicles.

    This project is undertaking the design of vehicle coordination protocols for mixed traffic environments with the objective of optimizing journey time predictability in both highway and urban settings. The project will investigate both centralised (i.e., availing of fixed infrastructure) and decentralised (self-organising) real-time coordination protocols capable of responding efficiently to perturbations in the traffic flow due to human driver behaviours and other incidents, as well as their integration with urban traffic control systems. The project will evaluate the benefit of such coordination under different AV penetration rates, in different driving environments, and with different levels of infrastructure provision.

  • Self-organizing Traffic System Optimisation

    The increasing availability of fine-grained sensor data, in particular, that expected to be available from individual vehicles as the level of automation increases towards fully autonomous vehicles, means that it will be possible to build an ever more detailed model of the current state of the road network, potentially including the intensions of drivers as derived from on-board navigation/routing systems. Coupled with the ability to exercise an increasing level of control over individual vehicles either indirectly, via urban traffic control and highway management systems, or directly, in case of autonomous vehicles or via advanced driver assistance systems (assuming that at least some proportion of drivers adhere to the advice), this offers the possibility of considering the deployment of new approaches to traffic management with the goal of optimising overall traffic system performance.

    This project is exploring the design of algorithms for optimisation of future urban/highway traffic management as a function of increasing levels of sensor data and increasing levels of control over individual vehicles. Such algorithms will need to take account of the scale, complexity, and inherently non-stationary nature of traffic systems, including accommodating the very many potential perturbations that can affect the system (e.g., pedestrians, weather events, unexpected obstacles/blockages, breakdowns etc.).

    Of particular interest are decentralised and/or self-organising algorithms that allow vehicles to converge towards a globally optimal solution over time. In doing so vehicles will need to learn how to collaborate with their neighbours in order to arrive at a mutually beneficial optimum that takes account of individual priorities as well as traffic control policies.

Support

Prof. Cahill's research is carried out with the support of Science Foundation Ireland and Intel.


Last updated 30 March 2022 by Vinny Cahill (Email).