Investigations
Automated Driving System (ADS)
NHTSA Preliminary Evaluation PE22014 — closed, opened 2022-12-12.
NHTSA investigation PE22014 is a Preliminary Evaluation opened on 2022-12-12 and currently closed. The subject is tracked inside the Office of Defects Investigation queue. Latest activity on this investigation was logged on 2024-08-20 — NHTSA updates that field whenever an Information Request goes out, a supplement is filed, or a status change is recorded in the public docket.
A Preliminary Evaluation like PE22014 is the entry point of the federal defect-investigation process. NHTSA engineers scan complaint databases, field reports, and manufacturer data to decide whether an Engineering Analysis is warranted, whether a voluntary recall is already sufficient, or whether the pattern does not rise to a defect finding.
Investigators summarized the matter as follows: "On December 12, 2022, the Office of Defects Investigation (ODI) opened a Preliminary Evaluation (PE22-014) to assess reports of inappropriately hard braking as well as immobilizations in vehicles operated by Cruise LLC (Cr..." Investigations are the early-warning layer of the federal auto-safety system, sitting upstream of formal recalls and defect orders. Whether this one closes without action or escalates into an Engineering Analysis, the full history stays in the ODI archive so researchers, litigators, and buyers can pull the paper trail at any time.
Investigation Summary
On December 12, 2022, the Office of Defects Investigation (ODI) opened a Preliminary Evaluation (PE22-014) to assess reports of inappropriately hard braking as well as immobilizations in vehicles operated by Cruise LLC (Cruise) and equipped with its Automated Driving System (ADS). Both inappropriately hard braking and vehicle immobilizations can increase the risk of a crash. To assess these risks, ODI collected information from Cruise, as well as from five peer ADS-equipped vehicle operators. Based on ODI’s analysis of immobilization data, the potential safety risk is heavily dependent on the context of the immobilized vehicle’s location, detectability of the vehicle due to hazard lights automatically engaging, and none of the immobilization incidents reviewed resulted in a crash or injuries. ODI also analyzed hard braking data which included 7,632 hard braking events commanded by the Cruise ADS. These events were determined by certain thresholds, including rate of deceleration, without regard to the appropriateness of the braking. ODI determined that Cruise vehicles contributed to 10 crashes that were reported under the Standing General Order 2021-01 (SGO), 4 of which involved a vulnerable road user and resulted in injury. On August 9, 2024, Cruise filed a safety recall (NHTSA Recall No. 24E-067) for “unexpected braking maneuvers [that] could occur if the ADS inaccurately predicts the immediate future path of the close-following actors, experiences diminished sensor precision from the close proximity of the close-following actors, or erroneously responds to a perceived risk ahead of the AV unrelated to the rear actor.” The recalled software versions affected all Cruise ADS-equipped vehicles within the US fleet. The Cruise ADS has been remedied through software updates that are intended to reduce the risk of unexpected braking maneuvers, including by improvements to perception, prediction, and planning. In view of the recall action taken by Cruise and ODI’s analysis of th
About This Investigation Type
A Preliminary Evaluation (PE) is the first phase of NHTSA's investigation process. It is opened when the agency identifies a potential safety defect pattern, usually triggered by consumer complaints, manufacturer reports, or field monitoring. During a PE, NHTSA gathers information to determine whether a formal engineering analysis is warranted.
Data from NHTSA Office of Defects Investigation. Cross-references: NHTSA recall campaign API and NHTSA FARS where fatality records overlap. PlainCars does not rate or recommend vehicles. Learn more.
Read our methodology — how this data is sourced, computed, and verified.