Aug 2023

Automated driving system (ADS) developers need a way to describe safe and competent driving for automated driving system-dedicated vehicles (ADS-DVs) in a way that is relatable to how stakeholders interpret safe driving today. The safety performance of some ADS behaviors can be measured and compared to naturalistic driving studies (NDS) data from human drivers to help characterize the socially acceptable balance between safety, lawful driving, efficiency, and comfort. This process enables developers to enhance the safety performance of ADS fleets by aligning it with human-relative benchmarks and considerations.

• ADS developers and manufacturers should establish context-relevant reference values and their uncertainties for safety performance metrics.
• Metrics employed in safety evaluations need to be complemented with reference values (and associated confidence intervals) that qualify the ADS performance as a comparison to such targets.
• This best practice outlines a process that can be followed to establish such targets, which, in turn, enable a clear evaluation of potential violations.
• Reference values should be commonly understandable to all stakeholders involved in the development and deployment of ADS systems, including regulatory bodies, developers, manufacturers, and end users.
• Naturalistic Driving Study (NDS) data of human drivers can be one input used to set these reference values.

July 2023

Successful scaling of automated driving system (ADS) technology and realization of its full potential will require developers and service providers to continuously monitor performance of their fleet and enact prompt improvements should issues arise. ADS developers and manufacturers can use the data collected from vehicles in active deployments (e.g., safety performance data) to proactively confirm initial risk assumptions and feed other safety management processes. Continuous monitoring of the automated driving system-dedicated vehicles (ADS-DVs) makes it possible to identify and address new risks related to changes in the environment. This also means the validity of operational design domain (ODD) related assumptions is periodically assessed.

• ADS developers and manufacturers should strive to continuously improve upon safety performance over time by reducing or preventing irregularities and strengthening confidence in design assumptions.
• Frameworks discussed in this document describe how to continuously monitor and potentially improve safety performance in the presence of known and unknown changes in the operating environment.

April 2023

Making risk management decisions such as safety and change management, safety analysis, and safety assurance are especially applicable when moving from concept to production intent for the ADS-DV. Change Risk Management (CRM) does not replace best practices or other methods for managing safety anomalies or change management processes.

• This AVSC information report provides a process for change risk management for fleet-operated ADS-DVs using level 4 or 5 automation.
• The processes and tools outlined are intended to help developers and manufacturers make determinations about risks not captured in the initial risk assessment and decide if further design and/or operational safety measures could be applied.
• Change risk management (CRM) is a process where ADS-DV manufacturers, developers, and fleet operators may evaluate if and how ADS-DV changes affect previously identified and managed safety risks. They may also identify, assess, and manage new safety risks resulting from those changes.

Aug 2022

VRUs pose challenges for the safe operation of ADS-DVs. Pedestrians, cyclists, micromobility users, and other VRUs take on different forms, movements, and behaviors. It is critical that developers rigorously test for safe operation when sharing the road with other types of road users.

•  It outlines types of VRUs and the challenges associated with detecting and classifying them. 
• This document also provides context for building behavioral competencies around safely responding to VRUs and applying metrics to evaluate those behavioral competencies.
• Finally, communication and properly calibrated expectations on the part of both ADS and VRUs will help ensure safe interactions.

Nov 2021

Managing the operational safety risks during testing is an important aspect of SAE Level 4 and 5 ADS development. A Safety Management System (SMS) is one approach designed to support organizational safety in a systematic and integrated way.

• Information Report shares information on a Safety Management System (SMS) framework considering ADS testing and evaluation.
• Create awareness of sound and effective approaches. Leverage approaches and/or frameworks used by other industries.
• Promote a safety culture, a system to assess and manage safety risk, evaluate risk control strategy effectiveness, and support organizational safety policies and objectives.

July 2021

There is value in assessing a set of behavioral competencies as directional indication of safety performance. Elemental behaviors can be used as starting point for additional assessment.

• Clarity on lexicon and relationship between maneuvers, object and event detection and response (OEDR), and behaviors.
• Glimpse into how AVSC members are using behavioral competencies within their organizations as part of a safety assurance framework.
• Elemental behavioral competency list can be used and customized by developers to demonstrate safety performance in whatever context they choose.
• Example of a repeatable method to tie in previous work of AVSC (ODD, Metrics) to develop application-specific metrics, using “Maintaining a Lane” to illustrate.

Mar 2021

Credible, practicable, and consistent ways to measure ADS performance are essential to garner public trust and confidence. Several ways have been put forward, but not widely adopted. Measuring safety outcomes is a complex task; requiring considerable time and exposure to achieve statistical significance.

• Foundational set of metrics to assess safety performance of the dynamic driving task.
• Recommends measuring safety outcomes (crashes and compliance with traffic regulations) and predictive metrics: maintaining a safety envelope, exhibit contextually safe vehicle motion control, and object and event detection and response (OEDR).
• Methods for assessing dynamic driving task (DDT) focused on aggregate results of several metrics

Dec 2020

Information for ADS developers on first responder awareness, understanding, standardized approach, and training for interacting with automated vehicles.

• Defines common first responder roles, interactions, and use cases.
• Recommended protocols for ADS developers and manufacturers to use when interacting with first responders across multiple use cases.
• Provides a framework to promote consistency for ADS developers and manufacturers in creating, distributing, and maintaining a first responder interaction plan

Sept 2020

Addresses data collection for event analysis for lessons learned from critical driving scenarios (i.e., what an ADS “saw” and the actions it took). Efficiently and consistently identifying safety issues for deployed vehicles is important to gaining the trust and confidence of the public.

• Common data collection practices for the purpose of event analysis and producing lessons learned.
• Clarifies areas specific to ADS-DVs.
• Definition of data elements, considerations for data prioritization, retrievability, survivability, storage, and traceability.

June 2020

No standards or industry guidance currently address passenger control (agency) for emergency reasons completely unrelated to the ADS’s ability to perform the Dynamic Driving Task (DDT).

• Recommended processes for passenger-initiated emergency
• Criteria and processes for passenger initiation of the features from inside a vehicle
• Communication with passengers, fleet operations, enhanced diagnoses of situation
• Interaction outside the vehicle with other road users, and general post-stop actions

April 2020

A description of an ADS Operational Design Domain (ODD), commonly defined terms and a framework in which to apply them are fundamental to promote consistent communication and help ensure that users’ ADS expectations are aligned with capabilities.

• A conceptual framework and lexicon for describing the ODD
• A consistent list of variables with definitions for describing the ODD

Nov 2019

In-vehicle fallback test drivers (IFTDs) contribute to safety during testing and therefore must be thoroughly trained on the automated driving system, the intended operational design domain, and controlling the vehicle to minimize risk.

• Qualifications and training for on-board human oversight of testing 
• Criteria and processes for selecting and training test personnel 
• Pre-work shift and post-work shift recommendations