Hardware
+ Software
  • Integrated Approach to Build the Full Stack Tech for Level-4 Self-driving Cars
  • XCU
  • xFusion
  • xUrban
  • xMap
  • xCloud
XCU
Our Vehicle Control Unit with a significant computation power to support Level-4 autonomy for R&D and deployment. Hardware synchronization among sensors, automotive grade cameras, and ASIL-D level safety island are all built in.
xFusion
Multi-sensor fusion technology to enable 360-degree perception with 300 meter range.
1
Far-range Lidar
5
Radar
2
Solid-state Lidar
6
GPS
3
Blind-spot Lidar
7
IMU
4
Autograde Camera
xUrban
The decision and motion planning engine that handles the most complex and dynamic traffic in urban cities in China.
xMap
Large-scale High Definition 3D Mapping system to map cities with Level-4 accuracy.
xCloud
Large-scale cloud-based platform to store, visualize, index, query, and connect the fleet of self-driving cars in the real-world and the virtual simulated worlds.

Data-driven Development Process

Data is the fuel of our success. AutoX has built a complete end-to-end toolchain, from hardware and software to the cloud, to leverage the significant amount of real-world and simulated data for the development of our self-driving system. AutoX is combining rigorous engineering, state-of-the-art machine learning, cloud computing, and big data to solve the challenge of our generation and reshape one of the world’s largest industries.
Data Visualization
xRay
We have built a complete data pipeline and cloud-based data service platform to store, compress, transfer, index, search, and visualize the significant amount of data at a petabyte scale. With friendly UI based on HTML5 and Node.js, xRay powers the rapid iteration of our development.
Large-scale Simulation
xSim
Inhouse purpose-built simulation engine supporting both log sim from real road test data and world sim from virtual scenario creation, and a seamless blending of both. Provides fast feedback to support R&D with regression tests and structured tests. Fully deterministic system to ensure reproducible and reliable simulation results. Integrated with AutoX XCU to support Hardware-In-Loop simulation.
HD Mapping Tool
xMap 2D
Inhouse pose graph optimization algorithm to ensure high definition mapping results at centimeter level accuracy. Not only with relative consistency between frames, but achieving absolute accuracy with global registration to latitude and longitude. Based on HTML5, we have built a complete toolchain for visualizing and annotating maps with rich semantic information, and supporting instant updates.
3D Mapping Tool
xMap 3D
Innovative technology for automatic traffic light annotation in 3D. Based on deep learning technology, we combine automatic traffic light detection results from multiple frames and aggregate in 3D space to generate 3D labels. This significantly increases the 3D mapping efficiency.

Designed for Volume Production

The first flat platform sensor mount design only 15cm thin, equiped with the most advanced automotive grade sensors
Solid-state Lidar
Radar
GPS
Far-range Lidar
Auto-grade Camera
Cleaning system
IMU
  • I have seen the future of mobility, and it is on the roads in Shenzhen!
  • Thomas Tsao, Cofounder and Chairman of Gobi Partners
  • Already at our first meeting in Silicon Valley,
    we were super impressed by the performance and
    maturity of AutoX’s AD stack.
  • Stefan Tilk, CEO of NEVS
Professor X
Founder
Founding Scientist of 3D Deep Learning
Solo Winner of Global MIT TR35 in Self-driving Industry Former Professor at Princeton University
Founder of the Princeton Computer Vision & Robotics Lab
PhD from MIT Computer Science and Artificial Intelligence Laboratory
Become X-men and X-women to make the world safer together!