Publicação
Leveraging vehicular communications in automatic VRUs accidents detection
| Resumo: | As technology advances on the field of Vehicular Ad hoc Networks (VANETs), there is a growing concern within the research community regarding the safety of the the Vulnerable Road Users (VRUs). These entities play an important role in traffic, but their typical agility and difficult to predict behavior pose challenges in the development of automatic systems that aim to protect them. The application of Machine Learning (ML) techniques on top of the communication data that can be collected from the road environment has the potential to predict VRUs movement, detect/locate them, or even compute probabilities of collisions. This paper proposes an automated and real-time VRU accident detection system (focused on motorcycles) by using neuronal networks with communication data that is generated by means of simulation, using the VEINS framework (coupling SUMO and ns-3). Results show that the proposed system is able to automatically detect any accidents between passenger vehicles and motorcycles at an intersection within 1 second, with an average of 0.61 second, after its occurrence. |
|---|---|
| Autores principais: | Ribeiro, Bruno |
| Outros Autores: | Nicolau, Maria João; Santos, Alexandre |
| Assunto: | Ciências Naturais::Ciências da Computação e da Informação |
| Ano: | 2022 |
| País: | Portugal |
| Tipo de documento: | comunicação em conferência |
| Tipo de acesso: | acesso restrito |
| Instituição associada: | Universidade do Minho |
| Idioma: | inglês |
| Origem: | RepositóriUM - Universidade do Minho |
| Resumo: | As technology advances on the field of Vehicular Ad hoc Networks (VANETs), there is a growing concern within the research community regarding the safety of the the Vulnerable Road Users (VRUs). These entities play an important role in traffic, but their typical agility and difficult to predict behavior pose challenges in the development of automatic systems that aim to protect them. The application of Machine Learning (ML) techniques on top of the communication data that can be collected from the road environment has the potential to predict VRUs movement, detect/locate them, or even compute probabilities of collisions. This paper proposes an automated and real-time VRU accident detection system (focused on motorcycles) by using neuronal networks with communication data that is generated by means of simulation, using the VEINS framework (coupling SUMO and ns-3). Results show that the proposed system is able to automatically detect any accidents between passenger vehicles and motorcycles at an intersection within 1 second, with an average of 0.61 second, after its occurrence. |
|---|