Labs & Research

Home > Labs & Research
Core Infrastructure: Sanya Supercomputing Center
The institute shares Geely's Hainan Sanya Supercomputing Center large-model training platform and million-scale GPU capacity, powering teaching and research.

20 PFLOPS

Total Compute (FP32)

Million-scale

GPU Compute Cluster

4–6 units

Dedicated Servers per Lab

24/7

Always Available

  • Supports Xinwei autonomous driving simulation cloud and large-model training platforms
  • Supports large-scale data anonymization and distributed model pruning & quantization
  • Undergraduates receive independent code repositories and simulation server access
  • All project progress syncs in real time via GitHub and internal Jira
  • Each lab server is equipped with four NVIDIA high-performance GPUs

Dedicated AI Laboratory Cluster

Program-aligned laboratory clusters provide advanced experimental teaching conditions:

Autonomous Driving Lab

Autonomous Driving Lab

Core Equipment

  • Autonomous driving simulation cloud platform
  • Drive-by-wire chassis experimental vehicle
  • Sensor calibration and fusion test bench
  • Multi-sensor data acquisition system
  • Long-tail scenario reproduction simulation environment

Capabilities

Supports end-to-end experiments from perception algorithm development to system integration testing, covering urban arterials, side roads, campus, and other scenarios.

Intelligent Cockpit Lab

Intelligent Cockpit Lab

Core Equipment

  • Immersive cabin simulator
  • Multimodal interaction test bench
  • Software flashing and diagnostics platform
  • In-vehicle display testing system
  • Intelligent voice interaction development kit

Capabilities

Supports multimodal HMI design and development, in-vehicle system integration testing, HMI prototype validation, and full-spectrum cabin experiments.

Research Directions

Research closely follows core intelligent vehicle technology needs, drawing on Geely Research Institute projects, national/provincial/municipal science programs, and industry–university–research collaborations.

Intelligent Perception
Vision-only perception, multi-sensor fusion, 4D mmWave radar, LiDAR point-cloud processing
Decision & Planning
Urban NOA core scenarios, end-to-end large models, dynamic programming, model predictive control
Safety Validation
Functional safety (ASIL), SOTIF, simulation validation platforms, long-tail scenario testing
Vehicle–Road Cooperation
V2X communications, vehicle–road–cloud integration, HD maps and localization, Robotaxi operations optimization

International Curriculum Recognition & Cooperation

Actively pursuing mutual recognition with leading institutions in Germany, the United States, and related fields—introducing advanced educational concepts and methods to elevate international talent development.

Quality Assurance System

Leveraging the Xinwei smart platform to track learning behavior in real time and monitor teaching quality. Project milestone reviews by joint university–enterprise expert panels ensure outcomes meet mass-production standards.

A long-term graduate competency tracking mechanism uses alumni feedback to iteratively refine programs and curricula.