Assessment of AEB algorithms and relevance of datasets used for AEB assessment
Typ
Examensarbete för masterexamen
Program
Automotive engineering (MPAUT), MSc
Publicerad
2019
Författare
Olleja, Pierluigi
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
Road fatalities are a major concern in modern society. The increase in the amount of
road vehicles on public roads exposes more people to the dangers that come with road
traffic. On highways, higher speeds are the cause of more serious crashes. To tackle
this issue, active safety systems work in cooperation with passive safety systems to
improve the occupants’ safety. In the assessment of new active safety systems, the
availability of crashes for computer simulation is limited. This work is therefore
aimed first at using lead-vehicle-braking events from the highD naturalistic driving
dataset as a basis for creating “what-if” crashes. These crashes were compared with
real crashes from an in-depth crash database (GIDAS) and an AEB algorithm was
assessed on both the datasets. The assessment was done using a common simulation
framework, developed in this work, and the results of crash avoidance and mitigation
were compared. Both datasets were collected on German highways. The results
showed comparable trends for the relative velocities involved in crashes from both
datasets, when comparing crashes where there was no reaction by the driver of the
following vehicle. The AEB assessment didn’t show clear similarities in crash
avoidance and mitigation.
Beskrivning
Ämne/nyckelord
Automated Emergency Braking (AEB) , naturalistic driving dataset , counterfactual simulations , rear-end collision , crash avoidance