The simulation platform has powerful scenario generation capabilities and is fully based on the OpenX standard. Thousands of simulation scenarios can be easily and efficiently generated in a variety of ways, effectively improving the authenticity, effectiveness and coverage of simulation tests, and escorting the mass production of automatic driving.
Dynamic Scenario Editor
Fully compatible with OpenSCENARIO 1.0 standard Support the placement and configuration of static and dynamic objects Support interactive editing of scenario elements such as Event, Action, and Condition Come with OpenSCENARIO text editor with syntax highlighting Support previewing the editing results of the debugging scenario in the case editor
Static Scenario Editor WorldEditor
Fully compatible with OpenDRIVE1.5 Support road and 3D scenario drawing of high-precision maps Preset various traffic element templates Automation tools to improve user editing efficiency Support the generation of high-precision maps through multiple data types Vector data file conversion Combine imagery, point cloud and 3D model sketch
Rich Scenario Generation Methods
Build the scenario through the scenario editor A toolchain based on semantic generalization enables fast definition of scenarios Convert into a scenario that can be used for simulation based on collected data
Provide the ability to generalize scenarios based on parameter combinations Semantic generalization toolchain provides the ability to quickly define and generalize scenarios
Third-Party Scenario Library
Scenario Library of Suzhou Automobile Research Institute of Tsinghua University Rich Dangerous (Edge) Scenarios Special thematic sub-library to meet the special testing needs of users China Automotive Research China typical scenario library v2.0 Standards and regulations, artificial experience data, traffic accident data in China, natural driving data
Outstanding sensor simulation ability and physical level real sensor simulation model make test verification more accurate.
Simulation of physical camera, lidar and millimeter wave radar based on real sensor calibration Support distributed multi-sensor clock synchronization and joint simulation Support hardware in the loop simulation of multiple sensors
Support 24-hour dynamic lighting Support rain, snow, fog dynamic weather system Built-in a large number of high-precision virtual assets Support automatic generation of 3D environment based on OpenDRIVE
Virtual Data Set
Support to generate RGB map, depth map, segmentation map, instance map and true value of object detection Support generating point cloud and true value of object detection
Planning and Control Simulation
Provide multiple simulation modules to easily meet various needs of regulatory testing.
Rich Data Interface
Support for a variety of common data interfaces, such as C++, Python, ROS, Simulink, etc Support various object level sensors, which can directly output target information Support multiple control modes, such as perfect control, brake throttle control, speed control, etc
Traffic Flow Simulation
Support the simulation of various traffic elements, including motor vehicles, non-motor vehicles, pedestrians, traffic lights, traffic rules, etc. Support vehicle following, lane change, intersection conflict model and signal light timing. Support the generation of random dynamic traffic flows in a parametric fashion Support third-party traffic simulation system co-simulation including SUMO, PTV Vissim
Built-in self-developed 26-degree-of-freedom dynamic simulation engine. Built-in dynamic simulation models of fuel vehicles and electric vehicles. Support the access of third-party dynamics modules such as CarSim®, CarMaker®, VI-grade®, etc. for co-simulation
Built-in driver model SimOne Driver, which can control the vehicle through control signals such as accelerator, brake, steering wheel angle, etc Support driver model and autonomous driving algorithm switching or hybrid control
There is a rich index library for users to choose and evaluate, covering multiple dimensions such as safety, violation, comfort, efficiency, economic energy consumption, and control accuracy Support users to flexibly configure the scoring function and evaluation weight Support multi-scenario concurrent evaluation Support single-scenario and multi-scenario statistical evaluation
Large-Scale Cloud Simulation Test
The native cloud architecture simulation platform is large-scale and highly concurrent, with a daily cumulative test mileage of 100,000 kilometers.
Cloud Native Architecture
Support large-scale concurrent simulation Support elastic expansion Flexible deployment to public cloud and private cloud Schedule cloud system based on K8S
Online Simulation Tool Chain Based on BS Architecture
Case management tools Scenario editing tools Main vehicle configuration tool Task management tools Simulation visualization tools, etc
Automatic Deployment CI / CD
Provide rich RESTful process control API Support version management of the algorithm via Docker hub Easy to be integrated into mainstream CI/CD frameworks
Cloud sharing of data to facilitate multi-role collaboration Support multi-user grouping and role authority management Support algorithm security and encryption management