The Project:
Perception for autonomous vehicles is critical for safety as it governs what happens in both prediction and path planning, and hence the trajectory taken by the vehicle. State of the art visual perception is provided by deep convolutional neural networks - black box machine learning models trained to perform a specific task on a chosen dataset. Their performance is usually evaluated against a benchmark using the same metric that the network was trained to optimise. How the performance of such a system transfers to real world scenarios or how its wider behaviour depends on conditions encountered in the real world is not well understood.
The project will focus on one particular scenario, and consider the sensing and perception requirements. In particular this will look at how capabilities of sensors can be modelled and how combinations of sensors can be used to provide confidence in the system capability. In terms of assurance the project would develop a systematic approach to assuring perception sub-systems. In practical terms, the project would develop perception sub-systems including object detection and classification algorithms and trial them on sub-scale vehicles, collecting evidence from trial operations to provide the evidence- base for the systematic approach to assurance allowing the approach to be validated. If practicable, the approach will be extended to full-size vehicles for further validation.
Research Supervision:
The successful candidate will conduct their research under the supervision of:
Prof John McDermid Director of the Assuring Autonomy International Programme (https://www.cs.york.ac.uk/people/jam)
Dr Will Smith, Reader in Vision Systems (https://www.cs.york.ac.uk/people/wsmith)
Dr Nick Pears, Reader in Computer Vision(https://www.cs.york.ac.uk/people/?username=nep).
Apply for this studentship:
You must apply online for a full-time PhD in Computer Science (https://www.cs.york.ac.uk/postgraduate/research-degrees/phd/#tab-4)
You must quote the project title (Safety of Perception for Autonomous Vehicles) in your application.
There is no need to write a full formal research proposal (2,000-3,000 words) in your application to study as this studentship is for a specific project.
You must also provide a personal statement of 500-1,00 words with your initial thoughts on the research topic.
Interviews are expected to take place within approximately 14 days of the closing date. The studentship will begin October 2021.
We will look favourably on applicants that can demonstrate knowledge of neural networks and image analysis.