CALVIN — a neural network that can learn to plan and navigate unknown environments | by Shu Ishida | Jun, 2022
Collision Avoidance Long-term Value Iteration NetworkSummary: CALVIN is a neural network that can plan, explore and navigate in novel 3D environments. It learns tasks such as solving mazes, by just learning fromvg expert demonstrations. Our work builds upon Value Iteration Networks (VIN) , a type of recurrent convolutional neural network. While VINs only work well in fully-known environments, CALVIN works even in unknown environments where the agent has to explore the environment in order to find the target.The three…