BROWSE
Detail
Conference Paper
딥러닝 기반 실시간 미래궤적 예측 시스템
Cited - time in 
Share 
Share
- Issue Date
- 2022-06
- Citation
- 대한전자공학회 학술 대회 (하계) 2022, pp.2601-2604
- Publisher
- 대한전자공학회
- Language
- Korean
- Type
- Conference Paper
- Abstract
- We introduce a real-time vehicle trajectory prediction system based on deep neural networks (DNNs). The proposed DNN for the trajectory prediction is specially designed to satisfy the following requirements: 1) efficiently utilize High-Definition (HD) map for the accurate prediction, 2) predict future trajectories of vehicles in a driving scene by taking their interaction into account, 3) predict multiple plausible trajectories for each vehicle, 4) predict future trajectories of vehicles in real-time. Due to the special design, the proposed prediction system can predict 10 trajectories per vehicle for 30 vehicles in the driving scene every 50㎳.
- KSP Keywords
- Accurate prediction, Deep neural network(DNN), High definition, Prediction System, Real-Time, Vehicle trajectory prediction