Alert button

What Can I Do Here? Learning New Skills by Imagining Visual Affordances

Jun 01, 2021
Alexander Khazatsky, Ashvin Nair, Daniel Jing, Sergey Levine

Figure 1 for What Can I Do Here? Learning New Skills by Imagining Visual Affordances
Figure 2 for What Can I Do Here? Learning New Skills by Imagining Visual Affordances
Figure 3 for What Can I Do Here? Learning New Skills by Imagining Visual Affordances
Figure 4 for What Can I Do Here? Learning New Skills by Imagining Visual Affordances

Share this with someone who'll enjoy it:

A generalist robot equipped with learned skills must be able to perform many tasks in many different environments. However, zero-shot generalization to new settings is not always possible. When the robot encounters a new environment or object, it may need to finetune some of its previously learned skills to accommodate this change. But crucially, previously learned behaviors and models should still be suitable to accelerate this relearning. In this paper, we aim to study how generative models of possible outcomes can allow a robot to learn visual representations of affordances, so that the robot can sample potentially possible outcomes in new situations, and then further train its policy to achieve those outcomes. In effect, prior data is used to learn what kinds of outcomes may be possible, such that when the robot encounters an unfamiliar setting, it can sample potential outcomes from its model, attempt to reach them, and thereby update both its skills and its outcome model. This approach, visuomotor affordance learning (VAL), can be used to train goal-conditioned policies that operate on raw image inputs, and can rapidly learn to manipulate new objects via our proposed affordance-directed exploration scheme. We show that VAL can utilize prior data to solve real-world tasks such drawer opening, grasping, and placing objects in new scenes with only five minutes of online experience in the new scene.

* 10 pages, 10 figures. To be presented at ICRA 2021  
View paper onarxiv icon

Share this with someone who'll enjoy it: