Advanced Deep Learning with Python

Model architecture

The following diagram illustrates the ChauffeurNet model architecture:

(a) ChauffeurNet architecture and (b) the memory updates over the iterations Source: https://arxiv.org/abs/1812.03079

First, we have FeatureNet (in the preceding diagram, this is marked by (a)). This is a CNN with residual connections, whose inputs are the top-down images we looked at in the Input and output representations section. The output of FeatureNet is a feature vector F which represents the synthesized network's understanding of the current environment. This vector serves as one of the inputs to the recurrent network AgentRNN, which predicts ...