Learning Long-term Visual Dynamics with Region Proposal Interaction Networks
This addresses the challenge of long-term predictions in visual dynamics for robotics and AI, offering a novel approach that generalizes to new environments.
The paper tackles the problem of learning long-term visual dynamics models by proposing Region Proposal Interaction Networks (RPIN), which reason about object trajectories in a latent feature space, resulting in significant improvements in prediction quality and planning ability over prior methods.
Learning long-term dynamics models is the key to understanding physical common sense. Most existing approaches on learning dynamics from visual input sidestep long-term predictions by resorting to rapid re-planning with short-term models. This not only requires such models to be super accurate but also limits them only to tasks where an agent can continuously obtain feedback and take action at each step until completion. In this paper, we aim to leverage the ideas from success stories in visual recognition tasks to build object representations that can capture inter-object and object-environment interactions over a long-range. To this end, we propose Region Proposal Interaction Networks (RPIN), which reason about each object's trajectory in a latent region-proposal feature space. Thanks to the simple yet effective object representation, our approach outperforms prior methods by a significant margin both in terms of prediction quality and their ability to plan for downstream tasks, and also generalize well to novel environments. Code, pre-trained models, and more visualization results are available at https://haozhi.io/RPIN.