Search papers, labs, and topics across Lattice.
This paper introduces a planning-based Inverse Reinforcement Learning (IRL) algorithm to learn world models from observation and interaction in real-world robotic manipulation tasks, addressing the challenge of learning without hand-designed rewards or demonstrator actions. The method learns image-based manipulation tasks from scratch in under an hour using only task observations. The learned world model representation demonstrates effective online transfer learning capabilities in real-world scenarios.
Robots can now learn manipulation tasks from scratch in under an hour using only visual observations and interaction, outperforming traditional IRL and RL methods in sample efficiency.
Observational learning requires an agent to learn to perform a task by referencing only observations of the performed task. This work investigates the equivalent setting in real-world robot learning where access to hand-designed rewards and demonstrator actions are not assumed. To address this data-constrained setting, this work presents a planning-based Inverse Reinforcement Learning (IRL) algorithm for world modeling from observation and interaction alone. Experiments conducted entirely in the real-world demonstrate that this paradigm is effective for learning image-based manipulation tasks from scratch in under an hour, without assuming prior knowledge, pre-training, or data of any kind beyond task observations. Moreover, this work demonstrates that the learned world model representation is capable of online transfer learning in the real-world from scratch. In comparison to existing approaches, including IRL, RL, and Behavior Cloning (BC), which have more restrictive assumptions, the proposed approach demonstrates significantly greater sample efficiency and success rates, enabling a practical path forward for online world modeling and planning from observation and interaction. Videos and more at: https://uwrobotlearning.github.io/mpail2/.
