Dynamic Scenario Representation Learning for Motion Forecasting With Heterogeneous Graph Convolutional Recurrent Networks.
Xing GaoXiaogang JiaYikang LiHongkai XiongPublished in: IEEE Robotics Autom. Lett. (2023)
Keyphrases
- recurrent networks
- learning process
- learning algorithm
- neural network
- recurrent neural networks
- learning systems
- dynamic environments
- supervised learning
- real world
- graph representation
- random walk
- motion estimation
- optical flow
- online learning
- multi modal
- image classification
- long term
- short term
- learning tasks
- feed forward
- biologically inspired
- prior knowledge
- deep learning
- multiple representations
- spatio temporal