Making Sense of Vision and Touch: Learning Multimodal Representations for Contact-Rich Tasks
This work addresses the challenge of sample-efficient robot control in unstructured environments, though it is incremental as it builds on existing self-supervised and reinforcement learning methods.
The paper tackled the problem of learning control policies for contact-rich manipulation tasks by using self-supervision to learn multimodal representations from vision and touch, improving sample efficiency and enabling generalization over varying conditions and robustness to perturbations in peg insertion tasks.
Contact-rich manipulation tasks in unstructured environments often require both haptic and visual feedback. It is non-trivial to manually design a robot controller that combines these modalities which have very different characteristics. While deep reinforcement learning has shown success in learning control policies for high-dimensional inputs, these algorithms are generally intractable to deploy on real robots due to sample complexity. In this work, we use self-supervision to learn a compact and multimodal representation of our sensory inputs, which can then be used to improve the sample efficiency of our policy learning. Evaluating our method on a peg insertion task, we show that it generalizes over varying geometries, configurations, and clearances, while being robust to external perturbations. We also systematically study different self-supervised learning objectives and representation learning architectures. Results are presented in simulation and on a physical robot.