Event
LEXP: Improving Robot Manipulation with Experiential Context Without Updating Weights
1 reports · 1 sources
Yihan Zeng and other authors propose LEXP, a robot manipulation framework that, under the condition that model weights and the action interface remain fixed throughout, combines task-level planning with an experience module to reuse execution experience to generate contextual feedback. The authors report across 12 real manipulation tasks: LEXP succeeded 23/36 times (63.9%), memory-only 19/36 (52.8%), and the original ChatGPT-6-Astra 2/36 (5.6%). On the three tasks of bread carrying, garbage and toiletries sorting, and drawer retrieval, removing planning lowered the success rate from 7/9 to 4/9, and removing retained experience lowered it to 3/9. All the above numbers are experimental results from the authors' paper.
Report timeline
- LEXP: Learning Robotic Manipulation from Physical Experience via Context-Space Adaptation
The authors propose LEXP, a framework for learning from physical experience through context-space adaptation that combines task-level planning with an experience module, reusing execution experience to generate contextual feedback while model weights and the action interface remain fixed. Across 12 real-world manipulation tasks, LEXP achieved 23/36 successes (63.9%), compared with 19/36 (52.8%) using memory alone and 2/36 (5.6%) for vanilla ChatGPT-6-Astra.
arXiv cs.RO ·
Related topics
Interest trend
- Current
- 3
- Peak
- 4Nov 4 01:00
- Past 24 hours
- –
The trend compares participants observed throughout the same period, so its range may be smaller than the current interest count. Move or tap to inspect each hour; use the arrow keys on a keyboard.