Pinlong Cai | 蔡品隆

cpl_pic.jpg

Research Scientist, Shanghai Artificial Intelligence Laboratory
Xuhui District, Shanghai, China
caipinlong@pjlab.org.cn
Google scholar | ORCID | ResearchGate

I have long focused on data-driven modeling for intelligent systems, applying these approaches to several domains such as Intelligent Transportation, Autonomous Driving, and Industrial Automation. Yet with the rapid emergence of Artificial General Intelligence (AGI), exemplified by large multimodal models, my perspective has gradually shifted from fitting tasks with data toward enabling machines to understand the world.

I believe true AGI should not merely be larger models, but systems that embrace knowledge-driven learning, perhaps by emulating human cognition, or through continuous self-evolution via interaction with the environment. Rather than replacing human intelligence, AGI will be an awakening of machine intelligence guided and accompanied by humanity. This evolution has the potential to fundamentally reshape how scientific discovery unfolds and profoundly advance the trajectory of human civilization.

I am committed to contributing to this journey, not just for the technology itself, but for the transformative future it may bring.


Work Experience

2021 - Now · Research Scientist · Shanghai Artificial Intelligence Laboratory
Knowledge Engine, Large Multimodal Model, Autonomous Driving
2020 - 2021 · Standard & Strategy Engineer · ZTE Corporation
V2X, Video Codec
2016 - 2017 · Research & Development Engineer · Quanzhou Institute of Equipment Manufacturing (CAS)
Computational Intelligence, Industrial Process Control

Education

2009 - 2020 · B.S. / M.S. / Ph.D. in Traffic Information Engineering and Control · Beihang University
supervised by Prof. Yunpeng Wang and Prof. Guangquan Lu

Achievements

Apr, 2026  · Heta (The multimodal knowledge engine for AI agents) has been released (Docs and Code)
Oct, 2025  · MUSE (Learning on the Job: An Experience-Driven Self-Evolving Agent for Long-Horizon Tasks) has been released (Code and Paper)
Sep, 2025  · HetaRAG (Hybrid Deep Retrieval-Augmented Generation across Heterogeneous Data Stores) has been released (Code and Paper)
Apr, 2024  · InternVL 1.5 (An open-source multimodal large language model) has been released (Code, Paper, Model and Dataset)
Apr, 2024  · LimSim++ (A closed-loop platform for deploying multimodal llms in autonomous driving) has been released (Code and Paper)
Feb, 2024  · OASim (An open and adaptive simulator based on neural rendering for autonomous driving) has been released (Code and Paper)
Dec, 2023 · Towards Knowledge-driven Autonomous Driving (Review Paper) has been released
Jul, 2023   · LimSim (A long-term interactive multi-scenario traffic simulator) has been released (Code and Paper)
Sep, 2022  · Data Compliance Guidelines for Intelligent Connected Vehicles have been released (Doc and Toolkit)

Academic Services

2020 - now · Journal reviewer of TPAMI, TKDE, TNNLS, TMM, TAI, ACM TIST, ACM TKDD, Info Fusion, Proc. IEEE, TITS, TR-C, TII, TVT, TCYB
2020 - now · Conference reviewer of CVPR, ICLR, IROS, ITSC, IVS, IAVVC, ICUS, TRB Annual Meeting
2024 - 2025 · Expert Member of Automotive AI Standard Project, Subcommittee on Intelligent & Connected Vehicles, SAC/TC114
Apr, 2024 · Invited talk at Hong Kong University of Science and Technology (Guangzhou), Empowering Autonomous Driving with LLMs/VLMs
Sep, 2023 · Invited talk at GOSIM Workshop, LimSim: A Long-duration, Interactive Multi-scenario Traffic Simulator
2020 - 2021 · Expert Member of Joint Video Experts Team (JVET), ITU-T SG 16 WP 3 & ISO/IEC JTC 1/SC 29
2020 - 2021 · Expert Member of Video Subgroup, Audio Video Coding Standard (AVS) Workgroup of China

Awards & Honors

Dec, 2025 · Shanghai Eastern Talent Program (Young Talent), Shanghai Government
Dec, 2022 · Shanghai Rising-Star Program, Shanghai Government
HTML Hit Counters