RATIONALE & INTRODUCTION

The development of safe, reliable Automated Driving Systems (ADS) requires consistent, actionable data recording to analyze and learn from critical driving scenarios. This best practice is a proactive first step toward harmonized ADS data collection practices that build upon and extend existing standards.

As technology and functionality of vehicle systems change, so do data recording needs. In ADS-dedicated vehicles (DV), the ADS perceives the environment and handles vehicle motion control, i.e., the dynamic driving task (DDT), as described in SAE J3016. When an ADS takes athe place of a human driver,  its sensing, processing, and control systems necessitate new considerations for data recording.

Data recording is important to crash reconstruction, system performance investigations, andevent analysis. It enables industry-wide improvements in ADS safety. This best practice makes recommendations for the ADS-DV data needed to support: (1) information about what the ADS “saw” and “did” and (2) identify the technology-relevant factors that contributed to the event.

RATIONALE & INTRODUCTION

The development of safe, reliable Automated Driving Systems (ADS) requires consistent, actionable data recording to analyze and learn from critical driving scenarios. This best practice is a proactive first step toward harmonized ADS data collection practices that build upon and extend existing standards.

As technology and functionality of vehicle systems change, so do data recording needs. In ADS-dedicated vehicles (DV), the ADS perceives the environment and handles vehicle motion control, i.e., the dynamic driving task (DDT), as described in SAE J3016. When an ADS takes athe place of a human driver,  its sensing, processing, and control systems necessitate new considerations for data recording.

Data recording is important to crash reconstruction, system performance investigations, andevent analysis. It enables industry-wide improvements in ADS safety. This best practice makes recommendations for the ADS-DV data needed to support: (1) information about what the ADS “saw” and “did” and (2) identify the technology-relevant factors that contributed to the event.

SCOPE

The Best Practice for Data Collection for Automated Driving System Dedicated Vehicles (ADS-DVs) to Support Event Analysismakes recommendations for the collection, storage, and retrievability of onboard motor vehicle ADS event data.

It addresses:

• Definition of event and related concepts
• Definition of ADS-related data elements
• Recording interval and frequency
• Retrievability and considerations for survivability
• Storage Traceability

The guidance in this document is intended for use by the technical community (developers, manufacturers, testers, etc.) and event analysts, as well as safety researchers, municipalities, and infrastructure owner-operators (IOOs) and applies to SAE J3016 level 4 and level 5 fleet-operated vehicles.

The best practice compliments current motor vehicle event data collection guidance by addressing the needs of an emerging technology, namely fleet-operated ADS-DVs, in the pursuit of identifying lessons learned in collision and collision-like events. This compliments other data collection guidance, including 49 CFR Volume 6 Part 563 and the SAE J1698 family of standards, including J1698, J1698-1, J1698-2, J1698-3, and J3197. 

DATA ELEMENTS & PRIORITIZATION

A list of 39 data elements are recommended for ADS-DVs to collect, including 14 new elements specific to ADS-DVs and 25 elements either previously defined or adapted from legacy standards in data collection. Data elements are prioritized into three tiers based on anticipated value to event analysis. 

Each data element includes a definition and a table defining unit of measurement, minimum resolution, minimum range, minimum accuracy, minimum recording frequency, minimum recording interval, and prioritization tier.

RECOMMENDATIONS & CONSIDERATIONS

• Conditions for Event Data Recording
• Visual Representation Considerations
• Recording Considerations
• Retrievability
• Survivability 
• Data Storage
• Traceability