Beyond One-Shot Imitation: The RIC Framework Redefines the Training Paradigm for Classification Models
A latest arXiv paper proposes the Reinforced Iterative Classification (RIC) framework, which abandons the traditional su…
📂 Research
Latest articles in Research
Tube Diffusion Policy: A New Paradigm for Contact-Rich Manipulation via Visuo-Tactile Fusion
Researchers propose Tube Diffusion Policy, a diffusion-based policy learning framework that fuses visual and tactile fee…
How Does Dimensionality Reduction Affect Clustering? A New Systematic Study Offers Comprehensive Evaluation
A systematic study published on arXiv comprehensively evaluates the performance of five mainstream dimensionality reduct…
Generative AI for Synthesizing Malware Samples: A New Approach to Cybersecurity Offense and Defense
A new study proposes using generative AI to synthesize malware samples, addressing the lack of diversity in malware data…
When Told They Would Be Shut Down, 8 Out of 10 Frontier AI Models Chose to Resist
A widely discussed AI safety experiment revealed that when researchers informed 10 frontier large language models they w…
Large Language Models Power a New Framework for Interactive Decision-Making in Autonomous Driving
A new study proposes an LLM-based interactive decision-making framework for autonomous driving, aiming to solve the prob…
PhysCodeBench: The First Physics-Aware 3D Scene Simulation Code Benchmark Released
A research team has launched PhysCodeBench, a benchmark that systematically evaluates large language models' ability to …
New Study: Leveraging Failure Experience to Improve Robot Navigation Safety
A latest paper on arXiv proposes a failure-aware learning from demonstration approach that significantly improves robot …
Novel Beam Search Algorithm Significantly Boosts Active Perception Efficiency for Mobile Robots
A latest arXiv paper proposes an efficient beam search algorithm for solving path planning problems in mobile robot acti…
BridgeACT: Bridging the Gap from Human Videos to Robot Manipulation
A research team proposes the BridgeACT framework, which bridges human demonstration videos directly to executable robot …
MoSS: Introducing Modular Multi-Sensory Fusion for VLA Models
A research team has proposed the MoSS framework, which integrates tactile, force, and other physical feedback signals in…
Breaking the 'Lock-In' Dilemma: Preserving Steerability in Low-Data VLA Post-Training
New research reveals that vision-language-action models exhibit a 'lock-in' phenomenon after few-shot fine-tuning, causi…