Learning to track environment state via predictive autoencoding
This work addresses the challenge of environment state estimation for applications like robotics or autonomous systems, but it appears incremental as it builds on existing concepts like Particle Filters.
The paper tackles the problem of learning forward models of stochastic environments from temporal unstructured image observations, enabling state tracking and long-term predictions. The model's performance is evaluated by comparing it to a Particle Filter in a simulated environment with moving objects.
This work introduces a neural architecture for learning forward models of stochastic environments. The task is achieved solely through learning from temporal unstructured observations in the form of images. Once trained, the model allows for tracking of the environment state in the presence of noise or with new percepts arriving intermittently. Additionally, the state estimate can be propagated in observation-blind mode, thus allowing for long-term predictions. The network can output both expectation over future observations and samples from belief distribution. The resulting functionalities are similar to those of a Particle Filter (PF). The architecture is evaluated in an environment where we simulate objects moving. As the forward and sensor models are available, we implement a PF to gauge the quality of the models learnt from the data.