# DiscoDemo: Using Reinforcement Learning to Generate Diverse Demonstration Data to Improve Imitation Learning

- Source: arXiv cs.RO
- Published: 2026-10-26T04:00:00.000Z
- RoboHot: https://robohot.news/en/items/u19yyzbj9krhideguzcrjicbp
- Original: https://arxiv.org/abs/2610.32093

## Summary

DiscoDemo uses a latent-conditioned reinforcement learning generator with privileged information, together with a parallel reverse curriculum and a success-gated diversity objective, to produce multiple successful closed-loop behaviors; after safety filtering, these are exported as a static dataset for downstream visuomotor imitation learning. On four manipulation tasks, the average downstream success rate in simulation rose from 40% for the strongest baseline to 62%, and on real robots from 39% to 78%. The authors believe that the diversity of successful solutions can improve imitation learning for pretrained flow-matching VLAs and complements corrective data collection.