Multi-Task Domain Adaptation for Deep Learning of Instance Grasping from Simulation
This work addresses the scalability of data collection for robotic manipulation, offering a domain adaptation solution for instance grasping, though it is incremental in leveraging existing simulation-to-real transfer techniques.
The paper tackles the problem of learning instance grasping in cluttered scenes by developing a multi-task domain adaptation framework that uses simulation data and transfers to real robots via domain-adversarial loss, achieving improved performance over baselines in real-world experiments.
Learning-based approaches to robotic manipulation are limited by the scalability of data collection and accessibility of labels. In this paper, we present a multi-task domain adaptation framework for instance grasping in cluttered scenes by utilizing simulated robot experiments. Our neural network takes monocular RGB images and the instance segmentation mask of a specified target object as inputs, and predicts the probability of successfully grasping the specified object for each candidate motor command. The proposed transfer learning framework trains a model for instance grasping in simulation and uses a domain-adversarial loss to transfer the trained model to real robots using indiscriminate grasping data, which is available both in simulation and the real world. We evaluate our model in real-world robot experiments, comparing it with alternative model architectures as well as an indiscriminate grasping baseline.