Professional Documents
Culture Documents
Inoa Ea22002 3184
Inoa Ea22002 3184
Investigation: EA 22-002
Prompted by: PE21-020
Date Opened: 06/08/2022
Investigator: Steven Posada Reviewer: Gregory Magno
Approver: Stephen Ridella
Subject: Autopilot & First Responder Scenes
PE21-020 is upgraded to an Engineering Analysis (EA) to extend the existing crash analysis, evaluate additional data
sets, perform vehicle evaluations, and to explore the degree to which Autopilot and associated Tesla systems may
exacerbate human factors or behavioral safety risks by undermining the effectiveness of the driver’s supervision. In
doing so, NHTSA plans to continue its assessment of vehicle control authority, driver engagement technologies, and
related human factors considerations.
The attached Detailed Summary further describes NHTSA’s review to date and the basis for upgrade to an EA.
On August 13, 2021, NHTSA’s Office of Defects Investigation (ODI) opened a Preliminary Evaluation (PE21-
020) to assess the performance of Tesla’s Autopilot system (a system characterized by Tesla as an SAE
Level 2 driving automation system designed to support and assist the driver in performing the driving task)
available in Tesla vehicles. The investigation opening was motivated by an accumulation of crashes in which
Tesla vehicles, operating with Autopilot engaged, struck stationary in-road or roadside first responder vehicles
tending to pre-existing collision scenes. Upon opening the investigation, NHTSA indicated that the PE would
also evaluate additional similar circumstance crashes of Tesla vehicles operating with Autopilot engaged, as
well as assess the technologies and methods used to monitor, assist, and enforce the driver’s engagement
with the dynamic driving task during Autopilot operation.
Between August 31, 2021, and September 13, 2021, NHTSA sent information request (IR) letters to Tesla and
twelve other vehicle manufacturers, requesting production and field incident reporting data as well as
information concerning the engineering and performance of their systems designated as Level 2.
On October 12, 2021, NHTSA sent two additional sets of requests to Tesla: (1) an IR letter to obtain
information on the company’s changes to subject vehicles’ functionality through software updates intended to
improve the detection of emergency vehicle lights in low light conditions; and (2) a Special Order (SO) to
request information concerning Tesla’s use of nondisclosure agreements with consumers whose vehicles
were included in a Full Self-Driving (FSD) “beta” release program.
NHTSA augmented its PE21-020 analysis with crashes reported by Tesla in response to an IR letter sent on
April 19, 2021, and via the Standing General Order (SGO) issued by NHTSA on June 29, 2021 (amended on
August 5, 2021).
Collectively, these sources provide NHTSA with a significant number of complaints and crashes involving
vehicles equipped with systems designated as Level 2. Throughout the course of this investigation, NHTSA
has prioritized the collection and review of this information.
NHTSA undertook a detailed review of the crash pattern that formed the principal basis for opening PE21-020:
the eleven collisions of subject Tesla vehicles with other vehicles stopped at first responder scenes reported
between January 2018 and July 2021. During this investigation, six additional such incidents were
subsequently identified and added to the crash analysis: Three crashes (Orlando, Petaluma, Desert Center)
involved first responder or crash attenuator trucks in the roadway and occurred after PE21-020 opened. A
2020 additional crash (El Paso) in which a police vehicle was struck was reported via the SGO in March 2022.
Two additional 2021 crashes (Belmont and Mount Pleasant) involving a first responder and a crash attenuator
truck, respectively, were reported to NHTSA in response to NHTSA’s April 19, 2021, IR letter to Tesla. Further
review of the Laguna Beach crash listed on the PE21-020 opening resume has led to its removal from this
consideration because the struck vehicle was parked out of traffic with no lights illuminated. The Laguna
Beach incident remains within NHTSA’s broader crash analysis within this investigation.
The agency’s analysis of these sixteen subject first responder and road maintenance vehicle crashes
indicated that Forward Collision Warnings (FCW) activated in the majority of incidents immediately prior to
impact and that subsequent Automatic Emergency Braking (AEB) intervened in approximately half of the
collisions. On average in these crashes, Autopilot aborted vehicle control less than one second prior to the
first impact.
All subject crashes occurred on controlled-access highways. Where incident video was available, the
approach to the first responder scene would have been visible to the driver an average of 8 seconds leading
up to impact. Additional forensic data available for eleven of the collisions indicated that no drivers took
evasive action between 2-5 seconds prior to impact, and the vehicle reported all had their hands on the
steering wheel leading up to the impact. However, most drivers appeared to comply with the subject vehicle
driver engagement system as evidenced by the hands-on wheel detection and nine of eleven vehicles
exhibiting no driver engagement visual or chime alerts until the last minute preceding the collision (four of
these exhibited no visual or chime alerts at all during the final Autopilot use cycle).
During the PE, the agency also closely reviewed 191 crashes involving crash patterns not limited to the first
responder scenes that prompted the investigation opening. Each of these crashes involved a report of a Tesla
vehicle operating one of its Autopilot versions (Autopilot or Full-Self Driving, or associated Tesla features such
as Traffic-Aware Cruise Control, Autosteer, Navigate on Autopilot, and Auto Lane Change). These crashes
were identified from a variety of sources, such as IR responses, SGO reporting, SCI investigations, and Early
Warning Reporting (EWR). These incidents, which are a subset of the total crashes reported, were identified
for a particularly close review not only because sufficient data was available for these crashes to support a
detailed evaluation, but also because the crash scenarios appeared characteristic of broader patterns of
reported crashes or complaints in the full incident data.
A detailed review of these 191 crashes removed 85 crashes because of external factors, such as actions of
other vehicles, or the available information did not support a definitive assessment. As a primary factor, in
approximately half of the remaining 106 crashes, indications existed that the driver was insufficiently
responsive to the needs of the dynamic driving task (DDT) as evidenced by drivers either not intervening
when needed or intervening through ineffectual control inputs.
In approximately a quarter of the 106 crashes, the primary crash factor appeared to relate to the operation of
the system in an environment in which, according to the Tesla owner’s manual, system limitations may exist,
or conditions may interfere with the proper operation of Autopilot components. For example, operation on
roadways other than limited access highways, or operation while in low traction or visibility environments, such
as rain, snow, or ice.
For all versions of Autopilot and road types, detailed car log data and enough additional detail was available
for 43 of the 106 crashes. Of these, 37 indicated that the driver’s hands were on the steering wheel in the last
second prior to the collision.
Although this subset of crashes is not exhaustive, the crash review identified patterns in system performance
and associated driver behavior across different sets of circumstances that enable the agency to identify areas
of engineering inquiry that warrant an upgrade of this Preliminary Evaluation to an Engineering Analysis (EA).
With respect to driver behavior, during this PE, NHTSA examined information submitted by Tesla and peer
manufacturers in response to an IR question that requested driver engagement and attentiveness strategies
to the DDT during system operation designated as Level 2. Of those crashes involving first responder or
roadside maintenance vehicles for which car log data existed, under the driver engagement strategy alerts
were presented to only two of the drivers within 5 minutes of the crash. This suggests that drivers may be
compliant with the driver engagement strategy as designed.
A driver’s use or misuse of vehicle components, or operation of a vehicle in an unintended manner does not
necessarily preclude a system defect. This is particularly the case if the driver behavior in question is
foreseeable in light of the system’s design or operation. For systems labeled as SAE Level 2 ADAS, important
design considerations include the ways in which a driver may interact with the system or the foreseeable
ranges of driver behavior, whether intended or unintended, while such a system is in operation. This is
because these systems still depend upon the driver to maintain supervisory responsibility for the DDT,
whereas the vehicle features perform only a support role. As such, ensuring the system facilitates the driver’s
effective performance of this supervisory driving task presents an important safety consideration.
Accordingly, PE21-020 is upgraded to an Engineering Analysis to extend the existing crash analysis, evaluate
additional data sets, perform vehicle evaluations, and to explore the degree to which Autopilot and associated
Tesla systems may exacerbate human factors or behavioral safety risks by undermining the effectiveness of
the driver’s supervision. In doing so, NHTSA plans to continue its assessment of vehicle control authority,
driver engagement technologies, and related human factors considerations.