I work on multimodal perception, industrial embodied intelligence, robot learning, and AI systems for smart manufacturing. My research focuses on robust visual sensing, RGB-T / RGB-D fusion, UAV and anti-UAV perception, vision-language-action models, and deployable AI workflows for scientific and industrial scenarios.

I am trained in artificial intelligence, biomedical engineering, and mechanical engineering, with current research interests spanning intelligent integrated manufacturing, multimodal physical information fusion, industrial inspection, robotics, and AI for Science.

• Multimodal perception and sensor fusion
• RGB-T / infrared-visible object detection
• UAV and anti-UAV intelligent sensing
• Vision-language and vision-language-action models
• Robot learning and embodied intelligence
• Smart manufacturing, industrial inspection, and digital twins
• Physics-informed learning and AI for scientific / engineering systems
• Optical materials, NIR-II sensing, and AI-assisted characterization

• TianWen 天问 — plug vision-language models into your object detector in one config file · project page

• RA-CMCD — reliability-aware cross-model consensus distillation for small-target RGB-T anti-UAV detection; reproducibility artifact for a manuscript under review

• speculative-mechatronics — commitment policies for physical actions under LLM latency, on a MacBook-scale testbed
• voice-cmd-correction — cheap, robust correction of noisy voice commands for restricted-vocabulary robotics
• newton-vla-demo — a 3-minute classroom demo of embodied AI on a laptop: Newton physics + an LLM as the VLA brain
• robotic-manipulation-learning — chapter notes, derivations and runnable cases for robotic manipulation

• abb-offline-coder — offline AI assistant that turns Chinese task descriptions into ABB painting-robot RAPID programs · demo

• cv-multimodal-research-roadmap — research roadmap for multimodal learning and vision-language reasoning
• PhysicsSimResearch — survey notes on frontier physics simulation and intelligent sensing
• claude-skill — a curated collection of Claude Code skills

Ongoing work on RGB-T / RGB-D fusion, anti-UAV perception, robot VLA experiments, and manufacturing AI (defect generation, dynamic monitoring, process optimization, error compensation) lives in private repositories and will be released with the corresponding papers.

• Sensors and Actuators B: Chemical (2026, CAS Region 1) — Deep learning enabled magnetic/rare-earth hybrid nanorobots for multi-modal bioimaging and temperature sensing · co-first author · DOI
• Laser & Photonics Reviews (2025) — Dual-heterovalent codoping enables bright and rapid X-ray excited persistent luminescence for noise-free 3D imaging · DOI
• Journal of Mechanical Engineering (2025) — Physical information fusion of multi-modal data for online fault monitoring of selective laser melting integrated manufacturing

My current goal is to build a coherent research portfolio around:

Multimodal Perception + Industrial Embodied Intelligence + Robotics / VLA + Smart Manufacturing AI

The emphasis is on turning each repository into a reusable research asset with clear problem definition, reproducible experiments, polished documentation, and long-term scientific value.

• Personal site — https://hollis36.github.io/
• ORCID — 0009-0007-8422-3955
• Email — peifu@stu.xidian.edu.cn