Proprioceptive Encodings for Robust Robotic Manipulation
Explore how proprioceptive encodings improve robotic manipulation and zero-shot generalization in dynamic, real-world environments.
Target-Aligned Generation for Cross-Domain Offline RL
Explore Target-Aligned Coverage Expansion to bridge domain gaps and enhance policy transfer in cross-domain offline reinforcement learning.
Boost LLMs with Context Training & Active Info Seeking
Enhance large language models using context training with active information seeking for better adaptability and performance in specialized tasks.
Bridging Human and VLM Scene Perception Gaps with CSS
Discover how Counterfactual Semantic Saliency reveals key gaps between human and vision-language model scene perception for better AI alignment.
Detecting Specification Violations in AI Agent Skills
Discover how semantic fuzzing with Sefz uncovers specification violations in AI agent skills to enhance safety and trust in LLM-powered agents.