Malleable Agents for Re-Configurable Robotic Manipulators
This addresses the challenge of adapting agents to flexible robotic systems, but it is incremental as it builds on existing deep reinforcement learning and domain randomization methods.
The authors tackled the problem of designing a learning agent for re-configurable robotic arms with varying numbers of links, proposing a deep reinforcement learning agent with sequence neural networks and domain randomization, and demonstrated its ability to transfer and generalize efficiently in simulations on a 2D N-link arm.
Re-configurable robots have more utility and flexibility for many real-world tasks. Designing a learning agent to operate such robots requires adapting to different configurations. Here, we focus on robotic arms with multiple rigid links connected by joints. We propose a deep reinforcement learning agent with sequence neural networks embedded in the agent to adapt to robotic arms that have a varying number of links. Further, with the additional tool of domain randomization, this agent adapts to different configurations. We perform simulations on a 2D N-link arm to show the ability of our network to transfer and generalize efficiently.