DriveSafeAI
Lead Participant:
WAYVE TECHNOLOGIES LTD
Abstract
Wayve is building pioneering technology -- AV2.0 -- to power SAE Level 4 'no-user-in-charge' self driving. The AV2.0 autonomy stack uses end-to-end machine learning (ML), removing the need for HD maps or expensive sensors, in contrast to the traditional approach to building automated vehicles -- AV1.0\.
Project **DriveSafeML** brings together world-leading expertise in ML (Wayve) and ML safety validation (WMG) to solve this critical blocker in the commercialisation of ML-based automated vehicles services in the UK.
DriveSafeML will develop a novel Operational Design Domain (ODD) and behaviour-based approach for the validation of machine learning (ML) in Automated Driving Systems (ADSs) to ensure that the training and test scenarios used for ML development are exhaustive and representative of the deployment area of the ADS.
An Independent Advisory Committee will provide a range of end-user perspectives from across the AV supply chain to ensure the approach builds confidence and trust in AV2.0\. The outputs can also help to shape emerging regulatory and standardisation frameworks at national and UNECE levels by working closely with regulators and the UK Government through the CAVPASS programme to develop best practice and guidance on ML system safety validation.
Project **DriveSafeML** brings together world-leading expertise in ML (Wayve) and ML safety validation (WMG) to solve this critical blocker in the commercialisation of ML-based automated vehicles services in the UK.
DriveSafeML will develop a novel Operational Design Domain (ODD) and behaviour-based approach for the validation of machine learning (ML) in Automated Driving Systems (ADSs) to ensure that the training and test scenarios used for ML development are exhaustive and representative of the deployment area of the ADS.
An Independent Advisory Committee will provide a range of end-user perspectives from across the AV supply chain to ensure the approach builds confidence and trust in AV2.0\. The outputs can also help to shape emerging regulatory and standardisation frameworks at national and UNECE levels by working closely with regulators and the UK Government through the CAVPASS programme to develop best practice and guidance on ML system safety validation.
Lead Participant | Project Cost | Grant Offer |
---|---|---|
WAYVE TECHNOLOGIES LTD | £2,222,496 | £ 1,333,498 |
  | ||
Participant |
||
UNIVERSITY OF WARWICK | £569,262 | £ 569,262 |
INNOVATE UK |
People |
ORCID iD |
Daniel Quirke (Project Manager) |