Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2403.14648 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909145246990336 |
|---|---|
| author | Chen, Jiayu Joyce Shladover, Steven E. |
| author_facet | Chen, Jiayu Joyce Shladover, Steven E. |
| contents | As driverless automated driving systems (ADS) start to operate on public roads, there is an urgent need to understand how safely these systems are managing real-world traffic conditions. With data from the California Public Utilities Commission (CPUC) becoming available for Transportation Network Companies (TNCs) operating in California with and without human drivers, there is an initial basis for comparing ADS and human driving safety.
This paper analyzes the crash rates and characteristics for three types of driving: Uber ridesharing trips from the CPUC TNC Annual Report in 2020, supervised autonomous vehicles (AV) driving from the California Department of Motor Vehicles (DMV) between December 2020 and November 2022, driverless ADS pilot (testing) and deployment (revenue service) program from Waymo and Cruise between March 2022 and August 2023. All of the driving was done within the city of San Francisco, excluding freeways. The same geographical confinement allows for controlling the exposure to vulnerable road users, population density, speed limit, and other external factors such as weather and road conditions. The study finds that supervised AV has almost equivalent crashes per million miles (CPMM) as Uber human driving, the driverless Waymo AV has a lower CPMM, and the driverless Cruise AV has a higher CPMM than Uber human driving. The data samples are not yet large enough to support conclusions about whether the current automated systems are more or less safe than human-operated vehicles in the complex San Francisco urban environment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_14648 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Initial Indications of Safety of Driverless Automated Driving Systems Chen, Jiayu Joyce Shladover, Steven E. Computers and Society Systems and Control As driverless automated driving systems (ADS) start to operate on public roads, there is an urgent need to understand how safely these systems are managing real-world traffic conditions. With data from the California Public Utilities Commission (CPUC) becoming available for Transportation Network Companies (TNCs) operating in California with and without human drivers, there is an initial basis for comparing ADS and human driving safety. This paper analyzes the crash rates and characteristics for three types of driving: Uber ridesharing trips from the CPUC TNC Annual Report in 2020, supervised autonomous vehicles (AV) driving from the California Department of Motor Vehicles (DMV) between December 2020 and November 2022, driverless ADS pilot (testing) and deployment (revenue service) program from Waymo and Cruise between March 2022 and August 2023. All of the driving was done within the city of San Francisco, excluding freeways. The same geographical confinement allows for controlling the exposure to vulnerable road users, population density, speed limit, and other external factors such as weather and road conditions. The study finds that supervised AV has almost equivalent crashes per million miles (CPMM) as Uber human driving, the driverless Waymo AV has a lower CPMM, and the driverless Cruise AV has a higher CPMM than Uber human driving. The data samples are not yet large enough to support conclusions about whether the current automated systems are more or less safe than human-operated vehicles in the complex San Francisco urban environment. |
| title | Initial Indications of Safety of Driverless Automated Driving Systems |
| topic | Computers and Society Systems and Control |
| url | https://arxiv.org/abs/2403.14648 |