Vehicle Automation Safety

We aim to accelerate safer development and deployment of automated vehicles

There are two trends in the approach to develop automated vehicle technology:

  1. Vehicles which are designed for well defined locations (commonly referred to operational design domains), and

  2. Vehicles which should be capable to operate in undefined/unfamiliar locations.

Society needs confidence that automated vehicles will operate sufficiently safely, regardless of the preferred approach. Our automated vehicle safety research covers:

  • Place-based risk identification and definition of safety goals

  • Assessment of defined locations (Operational Design Domains) for safety cases

  • Definition of KPIs and metrics which describe and quantify safety

  • Automated vehicle behaviour analysis in context and acceptance criteria definition

  • System behaviour decomposition identifying risks with system/component technologies

  • Identification of test cases and quantification of test coverage

Safety

In preparation for the wider deployment of automated vehicles, Physical and Digital Infrastructure (PDI) considerations and requirements are important for both supervised and unsupervised autonomy.

Infrastructure degradation can be as simple as worn road markings.

PDI requirements

Nominal safety (also referred to as “safety of the intended function”) requires an understanding of both Expected and Acceptable behaviour. Alongside these behaviours, we use our Behaviours Matrix method to investigate potential undesirable behaviours and their repercussions. This approach supports the preparation or review of both a developer’s and operator’s safety case.

Our Behaviours Matrix method was developed with support from Innovate UK through the “CCAV - Advancing safely to full vehicle automation” programme, project no. 10015570.

Behaviours Matrix

Infrastructure challenges

Environmental challenges

What challenges and risks do the road layout and infrastructure design pose to the perception system on automated vehicles, and their capability to mitigate those challenges and risks? Where should geo-fencing be considered due to unsustainable risk?

What challenges and risks do the localised environmental conditions pose to the perception system on automated vehicles, and their capability to mitigate those challenges and risks? What environmental condition thresholds should be considered due to unsustainable risk?