Autonomous Driving

A full stack solution for intelligent driving based on digital twins, efficiently assisting customers in achieving mass production and market launch.

Digital Car
AVM
Automatic Parking

Digital Car

An intelligent driving full stack digital twin platform based on Unreal Engine

Efficiently validate autonomous driving algorithms, train and validate AI models in a hardware free environment, reduce real vehicle testing costs and risks, and accelerate the implementation of autonomous driving technology.

Our Advantages

Cost Efficiency

Shorten test cycles by 80%, reduce testing costs by 60%, and eliminate real-road testing risks.
Safety Assurance

Enable exhaustive regression testing with ultra-high coverage to safeguard autonomous driving deployment.
Technical Advantage

Power the evolution of advanced autonomous driving with efficient data generation and application solutions.

Virtual Scene

Scene construction: Builds editable HD scenes from maps, real-world data, or OpenX standards.
Scene generalization: Scales weather/lighting/obstacle types, materials, and motion trajectories.
Virtual Sensor

Physical grade sensor simulation can simulate the distortion and overexposure of real lenses.
Can simulate the physical characteristics of rain and fog attenuation, material reflection, etc. of LiDAR.
Virtual ECU

Accurate ECU simulation: Simulate the entire chain from chip instructions to bus communication.
EUC real-time monitoring: It can monitor the usage of hardware resources in real time and support fault injection testing.
Virtual Dynamics

High-Fidelity Dynamics Simulation:Models power transmission (engine/motor/gearbox/differential/tires) and body rotation/vibration from suspension forces.
Supports lateral/longitudinal/vertical control interfaces.
Data Factory

Sensor data generation: Supports generalization and fully automatic annotation of multiple sensor data generation.
Scene map data export: Generate OpenDrive map data based on the scene and accurately store scene road information.
MLOps

Automated training: Through the MLOps system, one click completion of model training, publishing, and full process visualization.
Automated evaluation: Based on multiple AI algorithm metrics, automate the evaluation of iterative algorithm model performance and mixed data performance.

Application Case

A Tier 1 Algorithm Project

A China Tier 1 supplier faced delays in parking algorithm development due to limited real-vehicle testing resources. Kotei’s DreamCarSim provided a 1:1 test scenario, integrating sensor, ECU, and vehicle dynamics models, and established HIL/SIL testing pipelines. The solution enabled concurrent testing with streamlined access to algorithm performance metrics, reducing testing time by over 80%.

AVM

Brand-New 540° HD Panoramic Imaging Solution

This brand-new 540° HD panoramic imaging solution significantly improves driving safety by providing a comprehensive view of the vehicle's surroundings.

Our Advantages

Safety Assurance

Blind-spot-free environmental awareness and enhanced safety assurance.
Efficiency

Immersive experience and improved operational efficiency.

Mass Production Across Multiple Platforms

Compatible with mainstream chip platforms such as NVIDIA, Qualcomm, and SemiDrive.
Compatible with QNX, LINUX, and ANDROID systems.
The average CPU usage of the SDK is less than 3 KDMIPS.
Ultimate Visual Experience

Supports fully automatic one-click stitching with no distortion or misalignment in the stitched areas.
The car model supports high-level rendering, delivering a cinematic visual experience.
The average stitching error is less than 2 pixels.
Comprehensive Safety Upgrades

Eliminates blind spots caused by A-pillars and rearview mirrors, reduces collision risks, and safeguards driving safety.
Adapts to complex road conditions, supporting special viewing angles such as wheel view and narrow road passage view.
Safety upgrades include features like fingertip parking, parking photography, and door opening warning.
Personalized Customization

Customization of radar wall and trajectory line effects.
Settings for dynamic blind spots and stitching seams.
Wheel hub view/floating window steering view/local zoom view/digital car wrap...

Application Case

Zeekr AVM

Leveraging the newly developed 540° High-Definition Surround View Monitoring (AVM) solution by Kotei Information, we provide customized product solutions for all Zeekr models, fully facilitating the rapid mass production and launch of Zeekr vehicles based on the Qualcomm 8295 cockpit platform.

Automatic Parking

Advanced automatic parking full stack solution

Based on the 5V12U sensor solution, advanced AI algorithms are used to automatically identify parking spaces and achieve one click parking, making parking easy and worry free.

Our Advantages

Safety

Balance driving comfort and safety.
Full-stack Autonomous

Full-stack autonomous software solutions.
Diverse Functions

Diverse functions to meet different user needs.

Multi-platform compatibility

High compatibility with multiple intelligent driving platforms, enabling rapid adaptation to mainstream chip platforms such as NVIDIA, Qualcomm, Horizon J6M, Black Sesame, and Weijing.
Full-Scene Automatic Parking

Supports full-scenario automatic parking functionality (parallel parking, perpendicular parking, diagonal parking, line parking, fusion parking, and spatial parking).
Cross-level Memory Parking

Supports cross-level memory parking functionality with a range of up to 2 km.
High Ai Perception Accuracy

Multi-scenario, multi-obstacle recognition with an accuracy rate of up to 95%.
Human-Like Path Planning

Supports multiple planning schemes, enabling rapid and precise human-like path planning with an average parking duration of 45 seconds.
High Control Accuracy

Lateral and longitudinal accuracy of ±5 cm.

Application Case

Tier 1 Automatic Parking Project

A domestic Tier 1 automatic parking domain control project based on the 5R10V12U platform implements multiple high-level functions such as automatic parking, remote control parking, and memory parking. The perception, positioning, planning, and parking algorithms are all self-developed, with a parking performance pass rate of up to 95%.

Pioneering the AI-Defined Automotive Era

Let's Start Something Great Together