Self-Supervised Learning of Multi-Object Keypoints for Robotic Manipulation
This addresses the challenge of learning policies from raw camera observations in real-world robotic manipulation, where ground truth scene features are unavailable, though it appears incremental as it extends prior work to multi-object scenes.
The paper tackles the problem of computationally expensive and data-hungry policy learning in robotic manipulation by proposing a self-supervised method to learn multi-object keypoints via a Dense Correspondence pretext task. The result is a flexible and effective approach for sample-efficient policy learning, demonstrated on diverse robot manipulation tasks.
In recent years, policy learning methods using either reinforcement or imitation have made significant progress. However, both techniques still suffer from being computationally expensive and requiring large amounts of training data. This problem is especially prevalent in real-world robotic manipulation tasks, where access to ground truth scene features is not available and policies are instead learned from raw camera observations. In this paper, we demonstrate the efficacy of learning image keypoints via the Dense Correspondence pretext task for downstream policy learning. Extending prior work to challenging multi-object scenes, we show that our model can be trained to deal with important problems in representation learning, primarily scale-invariance and occlusion. We evaluate our approach on diverse robot manipulation tasks, compare it to other visual representation learning approaches, and demonstrate its flexibility and effectiveness for sample-efficient policy learning.