HeLM: Hierarchical Explicit Language Memory for Non-Markovian Manipulation
HeLM: Hierarchical Explicit Language Memory for Non-Markovian Manipulation
This research presents HeLM, a hierarchical Vision-Language-Action framework designed to handle non-Markovian manipulation tasks by maintaining task-relevant history through explicit language memory. Evaluated across 10 real-world tasks with 56 variants, including five newly formulated inter-episode tasks, HeLM improves intra-episode success by 12.8 percentage points over the strongest baseline and achieves 78% subtask success on compositional generalization tasks. The framework uses up to 4.5x less memory and runs 2.5x faster than the keyframe baseline at long horizons.
The paper introduces HeLM, a hierarchical VLA framework using explicit language memory to address non-Markovian manipulation challenges, evaluated on 10 real-world tasks with 56 variants.
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
memory reduction factor
≤4.5
View original evidence
M improves intra-episode success by 12.8 percentage points over the strongest baseline and achieves 78% subtask success on compositional generalization tasks, while using up to 4.5x less memory and running 2.5x faster than the keyframe baseline at long horizons.
Open source S4speedup factor
2.5
View original evidence
M improves intra-episode success by 12.8 percentage points over the strongest baseline and achieves 78% subtask success on compositional generalization tasks, while using up to 4.5x less memory and running 2.5x faster than the keyframe baseline at long horizons.
Open source S4number of task variants evaluated
56
View original evidence
inner monologue', we propose Hierarchical Explicit Language Memory (HeLM), a hierarchical VLA framework that maintains task-relevant history in explicit language through factorized event detection and event-triggered memory updates. We evaluate HeLM across 10 non-Markovian real-world tasks with 56 variants, including five newly formulated inter-episode tasks that require persistent context across task boundaries. HeL
Open source S3
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
M improves intra-episode success by 12.8 percentage points over the strongest baseline and achieves 78% subtask success on compositional generalization tasks, while using up to 4.5x less memory and running 2.5x faster than the keyframe baseline at long horizons.
Open source S4
inner monologue', we propose Hierarchical Explicit Language Memory (HeLM), a hierarchical VLA framework that maintains task-relevant history in explicit language through factorized event detection and event-triggered memory updates. We evaluate HeLM across 10 non-Markovian real-world tasks with 56 variants, including five newly formulated inter-episode tasks that require persistent context across task boundaries. HeL
Open source S3
Source：arXiv Robotics — research abstracts · arxiv.org