PRIME: Scaffolding Manipulation Tasks With Behavior Primitives for Data-Efficient Imitation Learning
- 15 August 2024
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (10) , 8322-8329
Abstract
Imitation learning has shown great potential for enabling robots to acquire complex manipulation behaviors. However, these algorithms suffer from high sample complexity in long-horizon tasks, where compounding errors accumulate over the task horizons. We present PRIME ( PR imitive-based IM itation with data E fficiency), a behavior primitive-based framework designed for improving the data efficiency of imitation learning. PRIME scaffolds robot tasks by decomposing task demonstrations into primitive sequences, followed by learning a high-level control policy to sequence primitives through imitation learning. Our experiments demonstrate that PRIME achieves a significant performance improvement in multi-stage manipulation tasks, with 10–34% higher success rates in simulation over state-of-the-art baselines and 20–48% on physical hardware.Keywords
All Related Versions
Funding Information
- IIIS
- Tsinghua University
- Todor Stoyanov
- Wallenberg AI, Autonomous Systems, and Software Program
- National Science Foundation (EFRI-2318065, FRR-2145283)
- Office of Naval Research (N00014-22-1-2204)
- UT Good Systems
- Machine Learning Laboratory
This publication has 28 references indexed in Scilit:
- Augmenting Reinforcement Learning with Behavior Primitives for Diverse Manipulation Tasks2022
- Learning robust manipulation tasks involving contact using trajectory parameterized probabilistic principal component analysis2020
- RIDM: Reinforced Inverse Dynamics Modeling for Learning from a Single Observed DemonstrationIEEE Robotics and Automation Letters, 2020
- Efficient Bimanual Manipulation Using Learned Task Schemas2020
- Learning to combine primitive skills: A step towards versatile robotic manipulation §2020
- IRIS: Implicit Reinforcement without Interaction at Scale for Learning Control from Offline Robot Manipulation Data2020
- Neural Task Graphs: Generalizing to Unseen Tasks From a Single Video Demonstration2019
- Deep Imitation Learning for Complex Manipulation Tasks from Virtual Reality Teleoperation2018
- Supersizing self-supervision: Learning to grasp from 50K tries and 700 robot hours2016
- A constraint-based method for solving sequential manipulation planning problems2014