Help creating a Deep Q Network with variable input size

[RafaelaSeguro]

Hi! I am sing DJL to create a Deep Q-Network to be used for path planning. I have found an example that also implements a DQN and I am trying to adapt it to my environment and agent. The problem I am having is with defining the input of my neural network. The input in this case should be the state space, but the state space doesn't have a fixed size, as it grows while the agent explores the environment. In the example I have, the state space is fixed, so the networks are initialized and used with this fixed dimension. Can anyone help me understand if it possible to create a network with a variable input size and how to use it?
Thank you!