Interaction Models for Merging and Cut-in Scenarios
Sprache des Vortragstitels:
Englisch
Original Tagungtitel:
European Control Conference, ECC 2021
Sprache des Tagungstitel:
Englisch
Original Kurzfassung:
Experienced human drivers generally try to consider
not only the safety of their own vehicle but also to
avoid disturbing surrounding vehicles in a way that could
negatively affect the flow of traffic or even cause accidents. This
requires an estimation of the possible reaction of other traffic
participants. This paper addresses this kind of interaction
model and proposes qLPV models for two important scenarios,
merging and cut-in, which have high importance for safety
and traffic fluidity. The proposed models rely only on available
datasets, and sparse identification methods are used to identify
their parameters. Drone measurements from Germany and
China are used for identification and evaluation.