Waymo Details Criteria for Safe Deployment of Autonomous Vehicles
Waymo has released a new paper outlining the processes it follows to determine whether its autonomous driving technology, the Waymo Driver, is ready for deployment in real-world environments.
The publication provides an in-depth look at the twelve acceptance criteria that underpin its safety assessments, along with the governance structure used to ensure consistent decision-making.
Waymo provides autonomous ride hailing services
© Uber
The Waymo Driver is a self-driving system that is continuously refined and tested. Before it is introduced into new operational areas, platforms, or major software upgrades, Waymo applies a structured methodology to confirm its readiness. This process builds on the framework that supported the initial launch of its fully autonomous service in Phoenix, Arizona, in 2020.
Central to this methodology is the principle of “absence of unreasonable risk.” Waymo applies a combination of tools, including a Safety Framework, a Safety Case, and Safety Impact data, to ensure that its vehicles meet this benchmark.
Waymo’s Safety Framework sets out the company’s principles, methods, and organisational responsibilities related to safety. The Safety Case then evaluates how well this framework holds up under scrutiny, ensuring that theoretical assumptions align with real-world performance. Safety Impact refers to the ongoing review of operational data after deployment, helping to validate the original safety assessment.
This is the first time any autonomous vehicle (AV) company has publicly shared this level of detail regarding its safety processes
© Waymo
The company emphasises that identifying and mitigating rare or high-risk scenarios before deployment is a key requirement. Rather than relying on a single measure or outcome, Waymo uses a set of twelve criteria to make a well-rounded, evidence-led judgement on deployment readiness.
These criteria span a range of technical and operational areas, from system performance and robustness to risk mitigation and post-deployment monitoring. Each one contributes to an overall picture of readiness and informs the company’s broader Safety Case. Together, they allow Waymo to determine whether the residual risks (those that remain after mitigation) are sufficiently low to proceed with deployment.
This systematic approach supports decision-making across the community. Waymo’s internal governance bodies, including its Safety Framework Steering Committee and Safety Board, both help oversee and validate these deployment readiness decisions.
Waymo’s decision to publish its methodology aims to encourage greater transparency and contribute to the development of industry-wide standards for AV safety assessment. As of now, there is no universal approach to determining when an AV system is ready for deployment. By sharing its internal criteria, Waymo hopes to provide a practical reference point for others in the field, including regulators and policy-makers.
Mauricio Peña, Chief Safety Officer at Waymo said:
This paper advances the conversation on how companies can credibly assess and demonstrate the safety of autonomous driving systems. By being transparent about our process, we aim to support government officials and the public in understanding what ‘safe enough’ truly looks like for AV deployment.
The company also acknowledges that its framework will continue to evolve alongside its technology and services. Each new deployment environment brings different risks and operational demands, and the readiness assessment process is designed to adapt accordingly.