Towards Practical Multi-Object Manipulation using Relational Reinforcement Learning
This work addresses the challenge of data inefficiency in reinforcement learning for multi-object robotic manipulation, offering a practical solution for robotics applications.
The paper tackles the problem of learning robotic manipulation of multiple objects with sparse rewards by introducing a graph-based relational architecture that enables effective knowledge transfer from simpler to complex tasks, achieving data-efficient learning to stack six blocks into a tower and demonstrating zero-shot generalization to taller towers and new configurations like pyramids.
Learning robotic manipulation tasks using reinforcement learning with sparse rewards is currently impractical due to the outrageous data requirements. Many practical tasks require manipulation of multiple objects, and the complexity of such tasks increases with the number of objects. Learning from a curriculum of increasingly complex tasks appears to be a natural solution, but unfortunately, does not work for many scenarios. We hypothesize that the inability of the state-of-the-art algorithms to effectively utilize a task curriculum stems from the absence of inductive biases for transferring knowledge from simpler to complex tasks. We show that graph-based relational architectures overcome this limitation and enable learning of complex tasks when provided with a simple curriculum of tasks with increasing numbers of objects. We demonstrate the utility of our framework on a simulated block stacking task. Starting from scratch, our agent learns to stack six blocks into a tower. Despite using step-wise sparse rewards, our method is orders of magnitude more data-efficient and outperforms the existing state-of-the-art method that utilizes human demonstrations. Furthermore, the learned policy exhibits zero-shot generalization, successfully stacking blocks into taller towers and previously unseen configurations such as pyramids, without any further training.