From Real World to Logic and Back: Learning Generalizable Relational Concepts For Long Horizon Robot Planning

Naman Shah, Jayesh Nagpal, Siddharth Srivastava

Keywords: Learnng symbolic abstractions, Symbolic world model learning, Learning for task and motion planning, learning for planning

TL;DR: We present the first known approach for learning generalizable and transferrable symbolic representations from unsegmented and unlabeled raw tracjtories.

Abstract:

Humans efficiently generalize from limited demonstrations, but robots still struggle to transfer learned knowledge to complex, unseen tasks with longer horizons and increased complexity.
We propose the first known method enabling robots to autonomously invent relational concepts directly from small sets of unannotated, unsegmented demonstrations. The learned symbolic concepts are grounded into logic-based world models, facilitating efficient zero-shot generalization to significantly more complex tasks. Empirical results demonstrate that our approach achieves performance comparable to hand-crafted models, successfully scaling execution horizons and handling up to 18 times more objects than seen in training, providing the first autonomous framework for learning transferable symbolic abstractions from raw robot trajectories.

Submission Number: 943