Bootstrapping Imitation Learning for Robotic Manipulation via Hierarchical Data Exploration Space
Authors:
- Jinrong YangJYJinrong Yang
- Kexun ChenKCKexun Chen
- Zhuoling LiZLZhuoling Li
- Shengkai WuSWShengkai Wu
- Liangliang RenLRLiangliang Ren
- Jing HaoJHJing Hao
- Wenqiu LuoWLWenqiu Luo
- Meiyu ZhiMZMeiyu Zhi
- Mingshan SunMSMingshan Sun
- Zhaojie ChuZCZhaojie Chu
- Fangping YangFYFangping Yang
- Yong ZhaoYZYong Zhao
Preprint
- 10 October 2026
- preprint
- Published by Elsevier in SSRN Electronic Journal
Abstract
Imitation learning (IL) is a powerful paradigm for complex, long-horizon robot manipulation, yet standard IL suffers from covariate shift: small, compounding errors during autonomous execution drift the agent away from the expert's demonstration distribution, leading to task failure. This brittleness stems from data collection that relies on monolithic, repetitive demonstrations, producing datasets that are highly redundant and intolerant to minor perturbations. We propose Hierarchical Data Exploration Space (Hide-Space), a data collection framework that builds an inherently robust training distribution to preemptively mitigate covariate shift. Hide-Space decomposes long-horizon tasks into atomic sub-tasks and collects data for each by actively exploring a continuous, overlapping state space that funnels towards a goal. Evaluations on four real-world long-horizon manipulation tasks show that IL policies trained on Hide-Space data achieve significantly enhanced performance with far fewer demonstrations, and offer insights into optimizing data quality and guiding data scaling.Keywords
This publication has 4 references indexed in Scilit:
- Behavior Transformers: Cloning k Modes with One Stone2022
- Imitation LearningACM Computing Surveys, 2017
- A survey of robot learning from demonstrationRobotics and Autonomous Systems, 2009
- Is imitation learning the route to humanoid robots?Trends in Cognitive Sciences, 1999