Jiang Chaokang
World Models · Generative Driving · 3D/4D Perception
World Models · Generative Driving · 3D/4D Perception
Hello, I'm Jiang Chaokang (蒋超康) — born in 1998 in Henan, China, and now a World Models Algorithm Engineer at Bosch China. I received my Master of Engineering in 2023 through a joint program (2020–2023) between the China University of Mining and Technology and Shanghai Jiao Tong University, where I was fortunate to be advised by Prof. Hesheng Wang and Prof. Yanzi Miao, and to build multi-year collaborations with Dr. Guangming Wang (Cambridge) and Dr. Jiuming Liu (Cambridge), among others. At the SJTU IRMV Lab I contributed to engineering projects such as a lawn-mowing robot and a 4D auto-labeling system. After graduating, I worked at PhiGent Robotics in Beijing (2023.06–2025.03).
Explore
Scene flow, LiDAR odometry, 3D/4D perception, world models, and generative driving.
A structured, filterable list where each card keeps the explanation short — one sentence for the problem and one for the core contribution — with verified PDF, arXiv, DOI, code, and project-page links.
Autonomous-driving, robotics, auto-labeling, deployment, and data-generation systems.
Each project is an independent, sanitized page covering context, role, the core algorithmic pipeline, and on-vehicle results — kept high-level for enterprise confidentiality.
Notes on models, systems, deployment, and research practice.
A personal blog for durable notes — model design, system trade-offs, deployment lessons, and engineering judgment captured over the long term.
Selected Publications
A streaming world model for autonomous-driving scenarios built on vector-graph diffusion flow.
Contribution. It aims to make future scene evolution modeling more efficient by operating on compact vectorized scene representations.
A large-scale 3D LiDAR point registration network with projection-aware transformer design.
Contribution. It improves registration efficiency and scalability by coupling 3D point registration with projection-aware 2D transformer modeling.
A journal version of diffusion-based uncertainty-aware 3D scene flow estimation.
Contribution. It extends the diffusion formulation into a hierarchical framework for more robust and uncertainty-aware point motion estimation.
Selected Projects
Built a Cosmos-Transfer2.5-based generative simulation platform: a 7V surround world model validated on internal data, real-map (Ingolstadt OSM → layout → 7V) scenario generation, a gRPC semantic bridge between WorldSim and the world model, the first 4-step distillation of 7V surround video (rCM + DMD2) for up to ~13.9× speedup, an editable platform for rare interaction data, and an all-in-one OneModel that serves layout generation, Gaussian-Splatting fix, and harmonization from a single denoiser.
Built a two-level controllable driving simulator: a structure-aware temporal vector VAE (STAR-AE) that compresses sparse, variable agents and lanes into fixed latents, a conditional latent-diffusion generator (STRIDENet) that produces history-consistent future traffic, and a sensor-level closed-loop WorldSim that fuses Gaussian-Splatting reconstruction, traffic-flow generation, and a mask-guided DiT video editor (built on MagicDrive-V2) into photorealistic surround rollouts.
Built a controllable surround-view driving generator that compresses 3D boxes and maps into spatial conditions, encodes text / reference frames / lanes / camera calibration into condition tokens, and injects them into a UNet diffusion backbone — producing cross-camera-consistent 4V / 7V / 11V images and video for data augmentation and open-loop simulation, evolving from OpenSora 1.0 + SD 3.5 to a MagicDrive-fused in-house model.
Type to search across publications, projects, and blog notes.