Autonomous driving simulation research: foreign giants build a closed-loop simulation platform, and domestic companies focus on scene library construction

In the development of high-level autonomous driving, the importance of simulation testing has become increasingly apparent. From 2020 to 2021, both foreign and domestic companies are stepping up their deployment in the field of autonomous driving simulation.

Foreign simulation giants are committed to building a closed-loop simulation platform and building industry barriers

In the process of advancing the market segment from industrial simulation to autonomous driving simulation, giants such as Ansys and Siemens have made up for the shortcomings of each MODULE of autonomous driving simulation through continuous acquisition, cooperation and technological development, and strengthened the closed loop of the simulation platform. Build industry barriers. At present, the layout of foreign leading companies has covered more than 80% of the simulation sector, and is continuing to update.

Figure: Automated driving simulation sector layout of some foreign simulation companies

Autonomous driving simulation research: foreign giants build a closed-loop simulation platform, and domestic companies focus on scene library construction

From 2020 to 2021, the updated and optimized sectors of foreign head simulation companies mainly include:

01. Support more sensor models such as camera, millimeter wave radar, lidar, etc.

Ansys: Add unified simulation of camera, lidar and millimeter wave radar

In February 2020, Ansys and FLIR System reached a cooperation to add FLIR thermal sensors to its driving simulator to realize the modeling, testing, and verification of thermal camera designs, and optimize functions such as AEB and pedestrian detection based on thermal cameras.

In January 2021, Ansys added the following functions in the latest 2021 R1 version: a new Sensor Viewer, which can help realize the 3D visual output of sensor detection truth information; a new Ansys VRX Sensor Simulator module, which can be directly installed in Ansys VRX The Driving Simulator platform implements camera, lidar and millimeter-wave radar simulation; the latest integrated full-physical millimeter-wave radar GPU accelerated simulation helps accelerate the simulation implementation.

Siemens: Enhanced Lidar Simulation

In September 2020, Siemens cooperated with XenomatiX, a supplier of solid-state lidars, and added a general-purpose, physically-based lidar simulation model to its simulation software Prescan. By comparing it with the actual road test data of XenomatiX, the simulation model can be fine-tuned. And optimization.

At present, the lidar model of Siemens Prescan platform can accurately describe all the optical properties of materials, including surface shape, reflection characteristics, material properties, and structure.

CARLA: Upgrade the LIDAR sensor

In September 2020, CARLA released version 0.9.10 to upgrade the lidar sensor. The new semantic-based LIDAR sensor can provide more information about the surrounding environment.

02, extend cloud simulation

Cloud computing capabilities can support large-scale simulation construction, cover a large number of driving scenarios, achieve high-concurrency running tests, and accelerate simulation running speed. Therefore, large-scale parallel acceleration in the cloud is one of the necessary core capabilities of the automated driving simulation test platform. Currently, the autonomous driving industry generally chooses Microsoft Azure as its cloud simulation partner.

For example:

In October 2020, Ansys cooperated with Microsoft to run Ansys VRXPERIENCE on Azure to improve scalability and provide cloud self-driving car simulation functions for customers of both parties.

In December 2020, the AirSim version was updated, adding AirSim development environment deployment documents on Azure, and supporting cloud simulation.

In addition, Siemens PreScan has achieved compatibility with Microsoft Azure, Amazon AWS, and compatibility with some domestic cloud service providers and cloud platforms.

03. Co-simulation has become a key development direction

Different simulation platforms have different emphasis on simulation modules. For example, there are Carsim, CarMaker, VI-Grade, VeDYNA, etc., which focus on vehicle dynamics simulation, and Vissim, SUMO, etc., which focus on traffic flow simulation. From 2020 to 2021, more and more solution providers will open software interfaces and conduct co-simulation with other solutions.

In January 2020, the new version of Prescan supports the generation of large-scale traffic flows. This function is supported by the Aimsun plug-in and the PTV Vissim plug-in. The Aimsun plug-in can support the synchronization of Aimsun’s microscopic traffic original scenes; the Vissim plug-in can perform co-simulation and data transmission between TASS PreScan software and PTV Vissim software.

In April 2020, CARLA released version 0.9.9, which can realize co-simulation with SUMO. In this mode, actions or events that occur in one simulator will be automatically propagated to another simulator.

In December 2020, CARLA released 0.9.11, which can realize co-simulation with CarSim, enabling users to create CarSim vehicles and control dozens of parameters including suspension systems and tires.

In January 2021, the CarSim 2021.0 version released by Mechanical Simulation, one of the highlights is the expansion of third-party software support. In the future, it will continue to open third-party software interfaces with technology partners such as Epic Games, NVIDIA, atlatec, Foretellix, monoDrive, Siemens, and CARLA.

Domestic companies start from the scene and win the ticket for autonomous driving simulation

The larger the coverage of the scene that the autonomous driving system can handle, the wider the area that the autonomous vehicle can drive. With the upgrade of autonomous driving functions, the number of scenarios that need to be tested and verified has increased geometrically during the upgrade iteration from L1→L4/L5, and the construction of the scenario library has become a vital part of the simulation industry chain.

At present, domestic companies such as Tencent, China Automobile Center, China Automobile Research Institute, Baidu, 51World and other companies have established their own autonomous driving simulation scene libraries.

Tencent: Can cover more than 99% of the simulation scenes of self-driving cars

Based on the powerful game engine capabilities, Tencent’s automated driving simulation platform TAD Sim has excellent performance in the realism and accuracy of the scene. Secondly, its scene generation system can derive 20 million virtual scenes based on 2000 types of logical scenes, covering more than 99% of autonomous vehicle simulation scenes.

Figure: Tencent TAD Sim 2.0: real data + game technology dual-engine drive

Autonomous driving simulation research: foreign giants build a closed-loop simulation platform, and domestic companies focus on scene library construction

In June 2020, Tencent released the TAD Sim 2.0 version. At the scene level, compared with 1.0, the new version breaks the barrier between real data and virtual data. Users can combine the traffic flow data collected by roads according to the needs of automated driving tests to form a virtual and physical test scene.

However, Tencent will not stop at autonomous driving. Its further layout is to further open its own capabilities, build virtual twin cities, and help smart cities and smart transportation. The specific approach is to integrate its game technology, cloud technology, and simulation technology to create a virtual twin platform to realize the mapping between the real world and the virtual world, and self-learn and update in the virtual world. In the future, it will also achieve iteration, prediction, and Ability to make decisions.

China Automobile Center: Add AD Scenario scene generator and expand the built-in scene library to 4,000

The built-in scene library of AD Chauffeur, a self-developed simulation platform of the China Automobile Center, covers natural driving scenes, standard and regulatory scenes, CIDAS dangerous accident scenes, and experience reorganization scenes.

In September 2020, the AD Chauffeur 2.0 version released by the China Automobile Center added the AD Scenario scene generator, which can solve the industry breakpoints such as scene construction, multi-source data format conversion OpenSCENARIO, logical scene splicing and reorganization, etc., providing for scene generation Important support. At the same time, it expanded the built-in scene library to 4,000 to meet the needs of users out of the box.

Picture: AD Chauffeur built-in scene library

Autonomous driving simulation research: foreign giants build a closed-loop simulation platform, and domestic companies focus on scene library construction

China Automobile Research Institute: Newly added typical accident scenarios, autonomous driving accident scenarios, and expected functional safety scenarios

In December 2020, China Automobile R & D released “i-VISTA China Typical Driving Scene Library V3.0”, adding typical accidents based on the previous version of natural driving scenes, standard and regulatory scenes, experience scenes and a small number of accident scenes. Scenarios, autonomous driving accident scenarios, anticipated functional safety scenarios, support automatic driving simulation test evaluation, and can be applied to simulation systems such as MIL, SIL, and HIL.

Picture: China Automobile Research Institute i-VISTA China Typical Driving Scene Library V3.0

In addition to the above-mentioned companies, Baidu relies on the advantages of Apollo technology to establish a simulation scene library; the “city-level all-element scene automation platform AES2021” released by 51World has improved the scene coverage and accuracy.

On the whole, China is still in a situation of fighting on its own in the construction of the scene library, facing problems such as inconsistent scene data, automatic driving simulation test evaluation, and difficulty in establishing a certification system. Therefore, the standard setting of the scene library has also become one of the current focuses in the simulation field.

2020-2021 Autonomous Driving Simulation Industry Chain Research Report (Part I) Contents

01. Introduction to Autonomous Driving Simulation

1.1 Overview of simulation technology

1.2 The significance of simulation testing for autonomous driving research and development

1.3 Types of Autonomous Driving Simulation Technology

1.4 Autonomous driving simulation software composition

1.5 International Organization for Standardization of Autonomous Driving Simulation

1.5.1 Introduction to ASAM

1.5.2 ASAM Recent Developments

1.5.3 C-ASAM Working Group

1.5.4 ASAM standard field and release route

1.5.5 ASAM’s OpenX series of standards

1.5.6 New ASAM standard for advanced autonomous driving

1.6 Current Status of China’s Autonomous Driving Simulation Test Standards

1.6.1 National Autonomous Driving Road Test Standard

1.6.2 Provincial and municipal autonomous driving road test standards

1.7 Status Quo of China’s Participation in the Development of International Standards for Autonomous Driving Test Scenarios

1.8 Autonomous driving simulation layout of OEMs

02. Autonomous driving simulation platform and company research

2.1 Typical components of an integrated simulation platform

2.2 Overview of the simulation software simulation system

2.3 Comparison of key foreign companies

2.4 Comparison of key domestic enterprises

2.5 ANSYS

2.5.1 Company Profile

2.5.2 ANSYS Product Distribution

2.5.3 ANSYS simulation capability enhancement

2.5.4 ANSYS investment acquisition event

2.5.5 ANSYS Autonomous Driving Simulation Tool Chain

2.5.6 ANSYS SCADE

2.5.7 ANSYS SCADE Suite

2.5.8 ANSYS VRXPERIENCE

2.5.9 ANSYS 2021 R1 update

2.5.10 Distribution of partners

2.5.11 Cooperation dynamics

2.5.12 ANSYS simulation planning

2.6 Siemens

2.6.1 Company Profile

2.6.2 Siemens solutions in the field of autonomous vehicles

2.6.3 Siemens’ expansion in ADAS and autonomous driving simulation

2.6.4 PreScan

2.6.5 Prescan sensor model

2.6.6 Sensor types and some scenarios supported by PreScan

2.6.7 PreScan automatic driving simulation and detailed functions

2.6.8 PreScan 2020 version update

2.6.9 PreScan application case

2.6.10 PAVE360

2.6.11 Supporting customers

2.6.12 Cooperation dynamics

2.7 NVIDIA

2.7.1 Company Profile

2.7.2 Simulation scheme

2.7.3 Drive Sim Software

2.7.4 Calculation module

2.7.5 DRIVE AV Safety Force Field

2.7.6 Cooperative events

2.8 CARLA

2.8.1 Introduction to CARLA

2.8.2 Basic structure of CARLA

2.8.3 CARLA product features

2.8.4 Carla 0.9.11

2.8.5 2020 version update

2.9 AVSimulation

2.9.1 Company Profile

2.9.2 Simulation software SCANeR

2.9.3 Simulation software SCANeR studio 1.9 version

2.9.4 Simulation software SCANeR studio version update

2.9.5 Cooperative events

2.10 LGSVL

2.10.1 Simulation scheme LGSVL

2.10.2 LGSVL version update

2.11 Panosim

2.11.1 Company profile

2.11.2 Simulation program PanoSim

2.12 AirSim

2.12.1 Open source simulation solution AirSim

2.12.2 AirSim version update

2.13 51World

2.13.1 Company profile

2.13.2 51Sim-One 1.3

2.13.3 51Sim-One version update

2.13.4 Cooperation events

2.14 Huawei

2.14.1 Company profile

2.14.2 Octopus, Huawei’s autonomous driving simulation platform

2.14.3 The combined ecology of cloud + AI + software and hardware + chips

2.14.4 Huawei Ascend 910 AI Chip

2.15 Baidu

2.15.1 Apollo simulation platform

2.15.2 Apollo Scene Library

2.15.3 AADS system

2.15.4 Cooperative events

2.16 Tencent

2.16.1 Tencent’s Autonomous Driving Layout

2.16.2 TAD Sim simulation platform

2.16.3 TAD Sim environment simulation

2.16.4 TAD Sim sensor simulation

2.16.5 TAD Sim version update

2.16.6 Derivative applications of autonomous driving simulation

2.16.7 Cooperative events

2.17 Ali

2.17.1 Alibaba Autopilot Layout

2.17.2 AutoDrive platform

2.17.3 Hybrid simulation test platform

2.18 BATH simulation business comparison

2.19 China Automobile Center

2.19.1 Company Profile

2.19.2 Driving scene simulation platform of China Automobile Center

2.19.3 AD Chauffeur2.0

2.20 China Automobile Research Institute

2.20.1 Company Profile

2.20.2 Comprehensive solution of scene library and simulation system

2.20.3 Virtual simulation scene library

2.20.4 Database of Typical Driving Scenes in China

2.20.5 Self-developed software tool chain

2.20.6 Data Processing Cloud Platform

2.20.7 Data Collection Platform

2.20.8 Cooperative events

2.21 Saimu Technology

2.21.1 Company Profile

2.21.2 Construction of the simulation test platform of Saimu Technology

03. Research on vehicle dynamics simulation

3.1 Research on vehicle dynamics simulation

3.2 Enterprise information related to vehicle dynamics

3.3 IPG Carmaker

3.3.1 IPG Automotive

3.3.2 CarMaker

3.3.3 Application of CarMaker in ADAS development

3.3.4 CarMaker customer application examples

3.3.5 CarMaker version update

3.3.6 Cooperative events

3.4 Carsim

3.4.1 Dynamics simulation software CarSim

3.4.2 Dynamics simulation software CarSim version update

3.5 AVL

3.5.1 Company Profile

3.5.2 AVL CRUISE

3.5.3 AVL Smart ADAS Analyzer

3.5.4 Cooperative events

3.6 Simpack

3.6.1 Product introduction

3.6.2 Simpack Automotive Application

3.6.3 SIMPACK version update

3.7 TESIS DYNAware

3.7.1 Introduction to TESIS Company and Products

3.7.2 TESIS DYNAware

3.7.3 DYNA4 simulation scenario

3.7.4 DYNA4 version update

3.8 MATLAB/Simulink

3.8.1 Product introduction

3.8.2 Automotive related solutions

3.8.3 Vehicle Dynamics Blockset

3.8.4 Automated Driving Toolbox

3.9 VI-Grade

3.9.1 Company Profile

3.9.2 VI-CarRealTime

3.9.3 DiM DYNAMIC Simulator

The Links:   NL6448BC26-09 FZ1200R33KF2C