Vision Transformer for Learning Driving Policies in Complex and Dynamic Environments

Tutkimustuotos: Artikkeli kirjassa/konferenssijulkaisussaConference article in proceedingsScientificvertaisarvioitu

4 Sitaatiot (Scopus)
73 Lataukset (Pure)

Abstrakti

Driving in a complex and dynamic urban environment is a difficult task that requires a complex decision policy. In order to make informed decisions, one needs to gain an understanding of the long-range context and the importance of other vehicles. In this work, we propose to use Vision Transformer (ViT) to learn a driving policy in urban settings with birds-eye-view (BEV) input images. The ViT network learns the global context of the scene more effectively than with earlier proposed Convolutional Neural Networks (ConvNets). Furthermore, ViT's attention mechanism helps to learn an attention map for the scene which allows the ego car to determine which surrounding cars are important to its next decision. We demonstrate that a DQN agent with a ViT backbone outperforms baseline algorithms with ConvNet backbones pre-trained in various ways. In particular, the proposed method helps reinforcement learning algorithms to learn faster, with increased performance and less data than baselines.

AlkuperäiskieliEnglanti
Otsikko2022 IEEE Intelligent Vehicles Symposium, IV 2022
KustantajaIEEE
Sivut1558-1564
Sivumäärä7
ISBN (elektroninen)978-1-6654-8821-1
DOI - pysyväislinkit
TilaJulkaistu - 19 heinäk. 2022
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisussa
TapahtumaIEEE Intelligent Vehicles Symposium - Aachen, Saksa
Kesto: 5 kesäk. 20229 kesäk. 2022

Julkaisusarja

NimiIEEE Intelligent Vehicles Symposium, Proceedings
Vuosikerta2022-June

Conference

ConferenceIEEE Intelligent Vehicles Symposium
LyhennettäIV
Maa/AlueSaksa
KaupunkiAachen
Ajanjakso05/06/202209/06/2022

Sormenjälki

Sukella tutkimusaiheisiin 'Vision Transformer for Learning Driving Policies in Complex and Dynamic Environments'. Ne muodostavat yhdessä ainutlaatuisen sormenjäljen.

Siteeraa tätä