Hidden Parameter Recurrent State Space Models For Changing Dynamics Scenarios
This addresses the challenge of modeling changing dynamics in control applications like robotics, though it appears incremental as it builds on existing RSSM frameworks.
The paper tackled the problem of recurrent state-space models assuming fixed dynamics, which is unrealistic for real-world scenarios like robotics, by introducing Hidden Parameter Recurrent State Space Models (HiP-RSSMs) that use latent factors to parametrize related dynamical systems, resulting in outperformance over RSSMs and multi-task models on robotic benchmarks.
Recurrent State-space models (RSSMs) are highly expressive models for learning patterns in time series data and system identification. However, these models assume that the dynamics are fixed and unchanging, which is rarely the case in real-world scenarios. Many control applications often exhibit tasks with similar but not identical dynamics which can be modeled as a latent variable. We introduce the Hidden Parameter Recurrent State Space Models (HiP-RSSMs), a framework that parametrizes a family of related dynamical systems with a low-dimensional set of latent factors. We present a simple and effective way of learning and performing inference over this Gaussian graphical model that avoids approximations like variational inference. We show that HiP-RSSMs outperforms RSSMs and competing multi-task models on several challenging robotic benchmarks both on real-world systems and simulations.