GPLAC: Generalizing Vision-Based Robotic Skills using Weakly Labeled Images
This addresses the challenge of achieving broad generalization in robotics without large interactive datasets, which is incremental by building on existing multitask and domain adaptation approaches.
The paper tackles the problem of learning robotic sensorimotor control policies that generalize to visually diverse and unseen environments by combining interaction data from a single environment with weakly labeled image data, resulting in greatly improved generalization as demonstrated in simulation and on a real robotic manipulator with substantial improvements over standard methods.
We tackle the problem of learning robotic sensorimotor control policies that can generalize to visually diverse and unseen environments. Achieving broad generalization typically requires large datasets, which are difficult to obtain for task-specific interactive processes such as reinforcement learning or learning from demonstration. However, much of the visual diversity in the world can be captured through passively collected datasets of images or videos. In our method, which we refer to as GPLAC (Generalized Policy Learning with Attentional Classifier), we use both interaction data and weakly labeled image data to augment the generalization capacity of sensorimotor policies. Our method combines multitask learning on action selection and an auxiliary binary classification objective, together with a convolutional neural network architecture that uses an attentional mechanism to avoid distractors. We show that pairing interaction data from just a single environment with a diverse dataset of weakly labeled data results in greatly improved generalization to unseen environments, and show that this generalization depends on both the auxiliary objective and the attentional architecture that we propose. We demonstrate our results in both simulation and on a real robotic manipulator, and demonstrate substantial improvement over standard convolutional architectures and domain adaptation methods.