React AI undertook a £250,000 research programme into aspects of connected and autonomous vehicles (CAVs), as part of the Flourish project.
Amongst other aspects of the project, we worked with the UK government’s Transport Systems Catapult (part of the Connected Places Catapult) to provide AI capability to an autonomous ‘pod’ vehicle, in colloboration with Airbus and the Bristol Robotics Laboratory.
This included collecting visual image data for training a novel neural network approach to route navigation
CAVs and Robotics
CAVs are a type of robotic system, and we saw Flourish as the perfect means for developing our software and applying it to the real world.
Access to the physical Pod was limited, and over the course of the project we developed a range of simulations and office-scale alternatives to continue our work.
Network-Level Machine Learning
To investigate the future impact of connected vehicles, the Flourish consortium has produced a micro-simulation of the Bristol region. Working with Atkins, this has been integrated with React AI’s cloud-based service to investigate the potential for neural networks to leverage the messages connected vehicles will produce, in order to produce accurate forecasts of traffic congestion.