I3L: Iterative Interactive Imitation Learning for Whole-Body Mobile Manipulation
I3L: Iterative Interactive Imitation Learning for Whole-Body Mobile Manipulation
I3L is an integrated hardware and learning system that addresses the reliability gap in offline imitation learning for whole-body mobile manipulation. It combines whole-body correction with iterative policy refinement, using a low-cost, open-source JoyLo+ interface and touch-triggered takeover. Evaluated across three robot embodiments and eight tasks, I3L improves autonomous success by 68.3 percentage points and achieves 90% success on all tasks, outperforming demonstration-only training by 35.6 percentage points.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
number of tasks
8
View original evidence
e robot embodiments and eight simulated and real-world tasks, including four long-horizon whole-body mobile manipulation tasks. Controlled simulation experiments show that four-round HG-DAgger achieves the highest final success on both tasks. Under this protocol, I3L improves autonomous success over the base policies by an average of 68.3 percentage points and achieves at least 90% success on every task, outperformin
Open source S4number of rounds
>4
View original evidence
g demonstration-only training by 35.6 percentage points on average with the same number of training trajectories. Averaged across tasks, takeover duration per episode decreases by 72% over four rounds. Together, these results show that I3L helps close the last-mile gap between offline training and reliable deployment and provides practical insights into collecting and using corrections for long-horizon whole-body mob
Open source S5
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
e robot embodiments and eight simulated and real-world tasks, including four long-horizon whole-body mobile manipulation tasks. Controlled simulation experiments show that four-round HG-DAgger achieves the highest final success on both tasks. Under this protocol, I3L improves autonomous success over the base policies by an average of 68.3 percentage points and achieves at least 90% success on every task, outperformin
Open source S4
g demonstration-only training by 35.6 percentage points on average with the same number of training trajectories. Averaged across tasks, takeover duration per episode decreases by 72% over four rounds. Together, these results show that I3L helps close the last-mile gap between offline training and reliable deployment and provides practical insights into collecting and using corrections for long-horizon whole-body mob
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org