Cruise Report by Quinn Emanuel Does not Provide Key Details to Regulators

---

Purpose

On a blog post from 1/25/2024 Cruise released a report they had commissioned by the law firm Quinn Emanuel Urquhart & Sullivan. The report also contained a technical analysis, though redacted, by Exponent Inc. The report was intended to be a full accounting to regulators, including the CPUC, of the October 2nd, 2023, crash where a pedestrian was purportedly dragged 20′ and seriously injured as a result. But a detailed reading of the report found it to not contain critical information. This omission of information to regulators is a concern because it shows that Cruise is still not taking adequate steps to concretely address safety.

The purpose of this article is to outline the main deficiencies of the Quinn Emanuel Report (QER). It is hoped that regulators will be able to use this information to improve the safety of autonomous vehicles moving forward, and to better regulate Cruise, in particular, more effectively.

The two key conclusions from this analysis is that the QER was deficient in stating that 1) the Cruise vehicle would have avoided striking and injuring the pedestrian if it had exhibited due care from the onset of the incident; and 2) the Cruise vehicle dragged the pedestrian for approximately 45 feet, not just the 20 feet that Cruise belatedly reported to regulators. In addition, the many other deficiencies of the QER listed here are also quite problematic.

Earlier Analyses of Crash

Prior to the release of the QER information, a detailed report on the October 2nd crash was written. Key problems were determined, including that the Cruise vehicle did not exhibit due care when approaching the pedestrian. Also, it was shown that the “pullover maneuver” actually could not be accomplished in the first place, indicating that it is not a valid reason for dragging the pedestrian 20 feet. These were examples of important information that was known to not have been presented to regulators, even before the QER was released.

In addition, on October 4th, less than two days after the crash, Michael Smith determined from evidence available that the pedestrian had been dragged 20 feet and seriously injured by the Cruise vehicle. Therefore Smith worked with Cyrus Farivar of Forbes to get that key information confirmed and publicized in Forbes on October 6th. This completely changed the narrative of the crash, even for the regulators. No longer could Cruise claim that all the fault of the pedestrian being injured were place on the human-driver of the other vehicle. And this information led to where we are today.

Deficiencies of Quinn Emanuel Report (QER)

The QER report is appreciated because it did provide sufficient new information, especially from the Exponent technical analysis that was included as an appendix.

This analysis is not about the needed operational changes that the QER recommends. Instead, it is about the details of what actually went wrong and how the regulators still have not been presented some of the most important details. A first draft of this analysis was emailed to the CPUC on January 26th, 2024.

Note: an annotated copy of the QER is provided at the end of this analysis. The copy is highlighted with comments that show where all of the information in this analysis came from.

Cruise vehicle did not exhibit due care from onset

Root cause of the Cruise vehicle hitting the pedestrian was that it did not follow Vehicle Code 21950 and exhibit due care and reduce speed when approaching a pedestrian. If the Cruise vehicle would have slowed down or stopped early, as required by law, it would have easily avoided colliding with the pedestrian in the first place.

California Vehicle Code 21950 is extremely clearly that a driver shall yield the right-of-way to a pedestrian, and that the driver shall exercise all due care and shall reduce the speed of the vehicle to safeguard the safety of the pedestrian:

California Vehicle Code 21950

(a) The driver of a vehicle shall yield the right-of-way to a pedestrian crossing the roadway within any marked crosswalk or within any unmarked crosswalk at an intersection, except as otherwise provided in this chapter.

(b) This section does not relieve a pedestrian from the duty of using due care for their safety. No pedestrian may suddenly leave a curb or other place of safety and walk or run into the path of a vehicle that is so close as to constitute an immediate hazard. No pedestrian may unnecessarily stop or delay traffic while in a marked or unmarked crosswalk.

(c) The driver of a vehicle approaching a pedestrian within any marked or unmarked crosswalk shall exercise all due care and shall reduce the speed of the vehicle or take any other action relating to the operation of the vehicle as necessary to safeguard the safety of the pedestrian.

(d) Subdivision (b) does not relieve a driver of a vehicle from the duty of exercising due care for the safety of any pedestrian within any marked crosswalk or within any unmarked crosswalk at an intersection.

(e)(1) A peace officer, as defined in Chapter 4.5 (commencing with Section 830) of Title 3 of Part 2 of the Penal Code, shall not stop a pedestrian for a violation of this section unless a reasonably careful person would realize there is an immediate danger of a collision with a moving vehicle or other device moving exclusively by human power.(2) This subdivision does not relieve a pedestrian from the duty of using due care for their safety.(3) This subdivision does not relieve a driver of a vehicle from the duty of exercising due care for the safety of any pedestrian within the roadway.

California Vehicle Code 21950

Following the vehicle code is not a technical issue so was not covered directly by the Exponent part of the analysis. But it is a key pedestrian safety issue and should have been fully addressed in the QER. And to point out how egregious the failure was of the Cruise vehicle to drive safely and follow the vehicle code, the Exponent data shows that the driving system actually continuously accelerated the vehicle while the pedestrian was in its path. Cruise not presenting this information on driving unsafely and violating the vehicle code to the CPUC is a serious dereliction of responsibility.

Note in the above chart from the Exponent technical analysis that the Cruise vehicle continued to accelerate (velocity increased) the entire time the pedestrian was crossing the lane of travel (-7.8s to -5.3s). And the vehicle continued at a high speed even after the pedestrian was detected to be hit by the other vehicle, the Nissan (-2.9s to -0.3s). In fact, the Cruise vehicle actually sped up more once the pedestrian started to depart the Nissan vehicle (-1.1 to -0.3s) towards the lane of the Cruise, a time of extreme danger and uncertainty. And even when the pedestrian had fallen into the travel lane in front of the Cruise vehicle, it still accelerated further (-0.8s to -0.5s). The velocity profile shows that instead of showing any kind of due care and slowing down, the Cruise vehicle actually drove in a dangerously aggressive, even belligerent way, making an injurious collision with the pedestrian unavoidable.

Pedestrian was dragged for approximately 45 feet, not just 20 feet

The pedestrian was dragged far further than the 20 feet reported. Not only did Cruise initially not adequately convey to the CPUC that the pedestrian was dragged, they only acknowledged that the pedestrian was dragged for 20 feet as part of the “pullover maneuver”. This was likely only admitted by Cruise because that is the distance reported by the October 6th Forbes article.

Page 31 of the QER states that the actual distance the Cruise vehicle moved during its “pullover maneuver” was approximately 7.5 meters, as stated by VP of Communications Aaron McLear. 7.5m is over 24 feet. Other estimates of 7 meters is still almost 23 feet. And the original information that leaked out from Cruise was that the distance was 22 feet. Therefore we will use a “pullover maneuver” distance of a very conservative 22 feet.

But there is another, even more important, factor. Details in Quinn analysis show that the Cruise vehicle was traveling quite fast, 18.4 mph, when it hit and ran over the pedestrian, and did not come to an initial stop for 1.8 seconds. The distance traveled during the 1.8 seconds is unfortunately redacted in the Exponent technical report. But based on a simple calculation, the distance it took for for the Cruise vehicle to stop after hitting the pedestrian would be roughly 23 feet. It therefore is certain that the pedestrian was dragged for a much longer distance than the 20 feet reported to the CPUC and public, a total of approximately 45 feet.

Being dragged this long distance of 45 feet clearly caused the traumatic injuries the pedestrian suffered. Not being explicit to the CPUC and other regulators about the actual distance the pedestrian was dragged is a serious misrepresentation to the CPUC that was not addressed by the submitted analysis.

Traumatic injuries of pedestrian not adequately addressed

The sole reason that Cruise was forced to stop operating and pursue the Quinn Emanuel Report (QER) was because a person suffered traumatic injuries. The key issue wasn’t about operations, coverups, nor technology. Yet the magnitude of the resulting injuries and the related communication of them was never addressed.

The QER misleadingly focused more on the pedestrian first being hit by the human-driven vehicle. The initial statement from Cruise was that the human-driven vehicle contact was “severe”, implying that it was the cause of the serious injuries. The QER should have pointed out that “severe” statement was not correct and appeared to be intended to mislead.

We now know that the pedestrian was hit by the Cruise vehicle when it was moving at 18.6 mph, quite a high speed. With hit by the Nissan vehicle the pedestrian went onto the hood and remained there for a short period of time. But the Cruise vehicle ran over the pedestrian, and then dragged them a total of approximately 45 feet. A trail of blood and skin were found where the pedestrian was dragged. The Cruise only made its final stop because while pinning down the pedestrian under a rear wheel the tire spun against the person, which would have been injurious itself. And then the pedestrian remained crushed under the Cruise vehicle at least until 9:40pm, for a minimum of 11 minutes. Keep in mind that the weight of a Cruise robotaxi is approximately two tons due to the very heavy 947 pound battery.

The resulting injuries caused the pedestrian to be hospitalized for two months, at least until November 28th, an indicating of severe and life altering injuries.

The QER should have directly addressed the human impact of the crash because it is the one thing we are working to prevent. Not doing so indicates that Cruise is still not serious about working with regulators to improve safety.

Emergency services not contacted by Cruise

The QER never addressed Cruise not contacting emergency services. A bystander being around and calling 911 was pure luck, and a robotaxi system cannot depend on the pure chance that there happened to be helpful people near the crash. Not having a system in place to quickly call local emergency services is a serious deficiency that should have been addressed.

Excessive response time

The QER never explicitly addressed Cruise personnel taking a very long time to show up at the scene of the serious injuries. The timeline on page 11 of the QER states “October 2, 10:17 p.m. Cruise contractors arrive at the Accident scene.” 10:17pm is 48 minutes after the crash time of 9:29pm. If the QER timeline is correct then the 48 minute response time of Cruise personnel to a serious injury crash is unacceptable and should have been noted explicitly.

Crash response protocol was not followed

QER failed to emphasize that Cruise did not follow their process for dealing with a serious injury incident. When originally submitting information to the CPUC to obtain their operating permit they surely included an incident response playbook to show the CPUC that they would handle situations properly. Them “aborting” and not following their response plan that their permit was partially based on is unacceptable and a serious misrepresentation to the CPUC.

Cruise vehicle hit pedestrian at much higher speed than implied

Cruise had initially reported that the vehicle “braked aggressively” before hitting the pedestrian, and “stopped immediately”. The details in the analysis shows that this was a gross exaggeration since the vehicle only braked for 0.25 seconds and it just slowed from 19.1 mph to 18.6 mph before hitting the pedestrian. That small amount of braking could not have materially lessened the injuries. The QER should have clarified that Cruise using the term “braked aggressively” was a significant misrepresentation, apparently intended to deflect blame for the serious injuries to the human-driven vehicle.

On page 31 of the QER, a Slack message from VP of Communications Aaron McLear makes it absolutely clear that releasing certain parts of the video would contradict the false narrative that the Cruise “braked aggressively”:

And on page 32 the QER provides details on how Cruise pushed the false narrative that the vehicle came to a complete stop immediately, when actually it took 23 feet for the Cruise vehicle to stop after hitting the pedestrian. And the news media picked up and published this false narrative:

Cruise stating pedestrian was “launched” or “thrown” is misleading

The pedestrian was on top of the human-driven vehicle for almost two seconds. They were not instantly “launched” or “thrown” into the path of the Cruise vehicle, as Cruise has repeatedly and intentionally claimed, including to the CPUC. The QER should have stated that the phrases used by Cruise were very misleading, and apparently intentionally so.

On page 49 of the QER Managing Legal Counsel Andrew Rubenstein stated the following, indicating that “launched” was not the appropriate word to use:

Pullover maneuver could not and should not have been attempted

The analysis misleadingly minimized that conducting the “pullover maneuver” was based on bad map data. As pointed out in the initial robotaxi.rodeo analysis, the Cruise vehicle had no place to pull over to. Very clearly, there was a separated bike lane to the right, continuing for quite a ways.

The QER states that the reason it wrongly tried to pull over was because “the outermost lane edge was mislabeled in the semantic map.” It must be pointed out though that this is gross negligence. It was not a complication with their software. The problem was that their underlying street data, something they had years to collect, was not correct. And to compound the problem, their sensors must have been able to detect that there were soft hit posts delineating a separated bike lane. Even with their unacceptable problem with the underlying map data, the vehicle should have realized that it could not actually pull over and therefore should not move forward, which would have spared the pedestrian more serious injuries. 

Software fix was not for root cause of crash

Cruise, in their NHTSA software recall, stated it would prevent vehicles from moving forward even if it was a lateral collision. But the QER claimed that the root cause was the system not properly locating the pedestrian, as opposed to the issue being a lateral collision. The QER should have noted that the software recall done for the NHTSA was misleading if the QER analysis is correct.

But even more problematic is that the root cause of the pedestrian being hit was due to the Cruise vehicle not following Vehicle Code 21950 and exhibiting due caution.

Cruise focused on countering accurate news coverage

The QER describes in detail how the executives were supposedly overly focused on “rebutting the erroneous media narrative”, but this is a serious misrepresentation of the actual problem.

In light of an increasing barrage of negative press at a time when Cruise was seeking to
expand in San Francisco, certain senior Cruise executives focused on rebutting the erroneous
media narrative, using words like they were “under siege,” and “we have no fighting chance with these headlines/media stories…we are drowning.”

Because Cruise had the 14-second video within minutes of the Accident, Cruise knew that the media articles were inaccurate; the video showed clearly that another vehicle, the Nissan, had hit the pedestrian and fled the scene.

Page 24 of Quinn Emanuel Report

Note Figure 1 on the same page of the report and shown below. Figure 1 shows the news headlines referred to. Each of these headlines was completely accurate. A woman indeed was hit by a Cruise vehicle and suffered traumatic injuries. The media narrative was not erroneous. It did not need to be corrected.

The actual problem was that the Cruise executives were focused on deflecting the bad but accurate news that a Cruise robotaxi had hit and severely injured a pedestrian. And they were focused on deflecting the bad but accurate news instead of on investigating what went wrong. The QER unfortunately further reenforces the false Cruise story instead reporting to the regulators and public that Cruise executives were actively trying to hide the bad but accurate news.

QER continues the casting of all blame to the human-driven vehicle

The QER states “But for the human driver of the Nissan hitting the pedestrian, the October 2 Accident would not have occurred.” That is a serious misrepresentation to the CPUC. The question the CPUC is facing is why did the Cruise vehicle cause avoidable serious injuries, not what happened before the Cruise hit the pedestrian. Not being the first vehicle to strike the pedestrian does not absolve the Cruise vehicle of fault. There is no “Get Out of Jail Free Card” for when someone else first strikes a pedestrian. Striking and injuring a pedestrian must be avoided if possible. Otherwise the vehicle is at fault for any injuries or damage inflicted.

Operational deficiencies of remote operation

Missy Cummings published very useful commentary on how the serious deficiencies of Cruise’s remote operation were not adequately conveyed to the CPUC. The CPUC should expeditiously incorporate the findings from Cummings into their regulator strategy. The CPUC should also call out how the QER was deficient in providing this key information.

Crash, not Accident

The QER clearly shows that the creators were not versant with vehicle safety issues. It consistently inappropriately uses the term “accident” instead of “crash”. When we use the word “accident” rather than “crash,” we’re implying that a collision was somehow inevitable. We’re not only absolving the driver from responsibility but the streets as well. The Cruise vehicle could have avoided crashing into and seriously injuring the pedestrian simply by exhibiting due care and slowing down at first sign of a problem. Therefore it was not an accident. It was a crash. The QER, if serious about vehicle safety, would have used the proper terminology.

Cruise did not proactively inform regulators nor the public

Throughout the QER there are examples of how Cruise leadership did not proactively communicate that the pedestrian was dragged during the “pullover maneuver”, but instead insisted that the video “speaks for itself”. But it is clear that the regulators, for various reasons, were not initially aware of the pedestrian being dragged. For the city officials meeting, they were asked to show the video 6 times, and then the audience still asked questions for confirmation. Clearly the video doesn’t speak for itself.

Senior Director of Federal Affairs, Eric Danko, clearly stated that they could suffice their obligation to regulators by only answering questions instead of being proactive:

And the CEO Karl Vogt even tried to minimize the evidence provide by any video:

And Cruise personnel were very happy with the success of their strategy of not proactively providing information. One employee at the DMV meeting apparently stated “they didn’t ask and we’re kind of lucky they didn’t ask” and also “phew, DMV didn’t notice” the dragging and “we dodged a bullet.”

The QER correctly calls out that Cruise should have proactively communicated all issues, and that not doing so was a failure. Instead, the Cruise strategy was clearly to not provide information unless they absolutely had to. And Cruise was happy with their initial success in limiting information to regulators. This is very relevant to the CPUC Ruling since it indicates that Cruise did indeed misinform the CPUC and that there should be significant consequences.

Good recommendations, but need to add Avoid those Problems in First Place

The QER provides excellent recommendations on what needs to change. It includes Organization Changes, Culture, Training, Continuous Improvement, Leadership, Process Improvements, Group Functional Improvements, and Reporting Approvals. But if Cruise simply replaces the leaders and pushes forward these changes, the original cause of the problems likely still remains. To fix things to need to address how Cruise got into this dysfunctional situation in the first place. Otherwise, with new people, the same culture and problems might simply be reestablished.

There must have been motivation for the Cruise culture to have developed the way it did. Likely it was due to dramatic changes in December 2021 when CEO Dan Ammann was pushed out by GM, the parent company. Most likely Ammann was pushed out because GM wanted to greatly accelerate the program. This meant that the subsequent leadership, including the new CEO Kyle Vogt, were under heavy pressure to move forward quickly, and start pulling in significant revenue. Aggressive goals were established, including $1B in revenue by 2025 (which now seems quaint). And Vogt was likely chosen as CEO because he had more of a “move fast (and break things)” style. If Cruise continues to have pressure from GM to ramp things up then the same problems will occur, even under new management. Therefore the most important issue is first establishing a proper relationship between GM and Cruise, or GM finding another owner.

QER doesn’t cover pedestrian safety issues called out by DMV

The DMV, and thereby the CPUC, revoked Cruise’s operating permits for two reasons, not just one. The reason for the DMV Suspension Order covered by the QER was with respect to problems communicating with the regulators, Cruise committing an act of omission.

The first reason for the DMV Suspension Order was with respect to safety, under 13 CCR § 227.42(c):

[T]he department shall immediately suspend or revoke the Manufacturer’s Testing Permit or a Manufacturer’s Testing Permit – Driverless Vehicles if a manufacturer is engaging in a practice in such a manner that immediate suspension is required for the safety of persons on a public road…Although the accident involved a vehicle operating under Cruise’s deployment permit, the behavior of the vehicle raises concerns that vehicles operated under Cruise’s driverless testing permit also lack the ability to respond in a safe and appropriate manner during incidents involving a pedestrian.

DMV Suspension Order

But the QER does not delve into what actually was done wrong. A prime issue is that the QER never covered that the Cruise vehicle did not follow Vehicle Code 21950 and exhibit due caution and slow down when the pedestrian was first at risk. If the Cruise vehicle was driven in a safe and reasonable way then it would have never hit the pedestrian. This should have been a key part of the QER.

Annotated Copy of the QER

To support this analysis of the QER, an annotated version was created with the key statements highlighted and commented on. This can also help readers of the long QER more quickly find the most relevant parts. The annotated version can be downloaded for viewing: