Unsupervised Learning of Visual 3D Keypoints for Control
This work addresses the challenge of sensorimotor control from high-dimensional images for robotics, offering a novel approach that improves performance over existing methods.
The paper tackles the problem of learning visual representations for robotic control by proposing an unsupervised framework to learn 3D geometric keypoints directly from images, which outperforms prior state-of-the-art methods across various reinforcement learning benchmarks.
Learning sensorimotor control policies from high-dimensional images crucially relies on the quality of the underlying visual representations. Prior works show that structured latent space such as visual keypoints often outperforms unstructured representations for robotic control. However, most of these representations, whether structured or unstructured are learned in a 2D space even though the control tasks are usually performed in a 3D environment. In this work, we propose a framework to learn such a 3D geometric structure directly from images in an end-to-end unsupervised manner. The input images are embedded into latent 3D keypoints via a differentiable encoder which is trained to optimize both a multi-view consistency loss and downstream task objective. These discovered 3D keypoints tend to meaningfully capture robot joints as well as object movements in a consistent manner across both time and 3D space. The proposed approach outperforms prior state-of-art methods across a variety of reinforcement learning benchmarks. Code and videos at https://buoyancy99.github.io/unsup-3d-keypoints/