Evaluating the Self-Learning Car in the Simulated and Real-World
The future self learning car must consider what features customers will want, what they will need and what features are worth automating. Perhaps most importantly we must consider how technology such as electric vehicles and self-driving cars will affect these user requirements. To this end, the direction of the research points towards the development of a future-proof framework for development of self-learning and autonomous features within vehicles.

Predicting Pedestrian Movement using external perception sensors
In order to safely navigate a complex environment human drivers will observe their surroundings and make predictions on likely future states. Human drivers are able to create relatively complex predictions of other humans behaviour based on a complex set of information including features such as the local geography, body pose, face direction etc. Anecdotally many drivers can attest to being able to tell whether a pedestrian is planning to cross in front of them or not just by looking at them. Autonomous or semi autonomous vehicles currently lack this complex prediction model, instead relying on simpler prediction models more prone to false positives. As higher level autonomous vehicles become more common it is important for them to be able to smoothly interact with pedestrians. As such this project aims to explore whether a machine learning algorithm, utilising automotive sensors, can be trained to produce a more accurate prediction model for pedestrian movement when compared with current human designed models.

A User-in-the-Loop Test Methodology for Wireless Network Services in Vehicles
The study will provide answers to industry’s nonstandard use cases such as how much complexity is added when investigating connected services for a moving car as opposed to a static car and how important are the user(s) activities for the in-car communication channel. The main objective of the project is to create a general model of the relationship between the wireless network parameters measured inside a moving car, possibly including its user interactions. This model will be useful as a prediction tool for assessing the perceived quality of in-car connected services.

Keeping the Driver Aware in a Semi Autonomous Car
What happens if a self driving car needs to hand back control to the driver, but the driver is distracted or unaware? This is a potentially challenging situation for the driver, vehicle and other road users alike. The research question is ‘How best can the car interact with the driver to keep them aware whilst the car is self-driving?’ This design challenge is the focus of this research, and is split into three key areas: Safety, Interfaces and testing.

On-Board and Off-Board Data
In the automotive industry, there is a strong trend toward an increased data output from the Electronic Control Units (ECU) in order to enable more complex features. This presents challenges that include: the bandwidth and topology constraints of the In-vehicle Network (IVN); the bandwidth and latency of wireless communication platforms (2G/3G/4G LTE) and the inability to effectively store, process and query the available information wherever it is needed. In order to make the most effective use of this data, this project aims to research techniques that can assist with data transfer and management between the On-Board and Off-Board systems.