AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Guided Distillation and Risk Adaptive Evolution for Multi-Robot Navigation

Xuyang Li, Jianwu Fang, Lin Li, Boyuan Chen, Guangliang Li, Jianru Xue

AAAI 2026 · Reinforcement learning › Visual and representation RL · approach: In-context and adaptive RL · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add PGM guidance to IQN trainingIQN · Maritime trainingmean reward: 0.8 vs. 0.45Improved
add PGM guidance to IQN trainingIQN · Maritime trainingcumulative successes: 50k vs. 20kImproved
set Pthres to 90th percentile instead of 95th (ablation)PGM-IQN · Maritimeaverage reward: about 0.69 -> about 0.80Improved
set Pthres to 90th instead of 80th/85th (ablation)PGM-IQN · Maritimeaverage reward: about 0.76 -> about 0.80Improved
add PGM guidance with any Pthres vs vanilla IQN (ablation)IQN · Maritimeaverage reward: about 0.46 -> about 0.80Improved
add PGM distillation without risk adaptation (ablation)IQN · Maritime TSR vs robot densityTSRImproved
add LLM-evolved risk-adaptive policy over Ada/ART variants (ablation)PGM-IQN · Maritime and TerrestrialTSRImproved
add LLM-evolved risk policy vs hand-crafted Ada (ablation)PGM-IQN · Maritime and TerrestrialCRMixed
GUIDER vs strongest baseline ART-IQNGUIDER · MaritimeTSR: 58.32% -> 72.20%Improved
GUIDER vs strongest baseline ART-IQNGUIDER · TerrestrialTSR: 76.50% -> 89.54%Improved
GUIDER vs Ada-IQNGUIDER · TerrestrialCR: 0.66% -> 1.67%Hurt
dual-confidence cascade with coarse fallbackPGM · PGM validationcoverage / accuracy: 96.98% accuracy for additional 17.91% of casesImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy