TACO: Learning Task Decomposition via Temporal Alignment for Control
This addresses the challenge of modular LfD for robotics and AI by enabling more efficient and generalizable task decomposition without heavy supervision, though it is incremental in improving existing methods.
The paper tackles the problem of decomposing complex tasks into simpler sub-tasks for Learning from Demonstration (LfD) by proposing a weakly supervised, domain-agnostic method based on task sketches, which reduces annotation effort and performs comparably to fully supervised approaches in domains like simulated 3D robot arm control.
Many advanced Learning from Demonstration (LfD) methods consider the decomposition of complex, real-world tasks into simpler sub-tasks. By reusing the corresponding sub-policies within and between tasks, they provide training data for each policy from different high-level tasks and compose them to perform novel ones. Existing approaches to modular LfD focus either on learning a single high-level task or depend on domain knowledge and temporal segmentation. In contrast, we propose a weakly supervised, domain-agnostic approach based on task sketches, which include only the sequence of sub-tasks performed in each demonstration. Our approach simultaneously aligns the sketches with the observed demonstrations and learns the required sub-policies. This improves generalisation in comparison to separate optimisation procedures. We evaluate the approach on multiple domains, including a simulated 3D robot arm control task using purely image-based observations. The results show that our approach performs commensurately with fully supervised approaches, while requiring significantly less annotation effort.