Understand automotive data collection, calibration and refeeding in one article
Latest update time：2023-10-18
Reads：
Author
: Wang Yingfei
Produced by : Automotive Electronics and Software
01.
What are data acquisition, calibration and refilling?
Automotive data collection, calibration and data recirculation are important steps in the vehicle development and testing process. They are
briefly
:
1. Automobile data collection:
Automobile data collection refers to the collection of data on the vehicle's surrounding environment, vehicle status, and driving behavior through sensors, cameras, radars and other equipment mounted on the vehicle. This data can include images, videos, lidar
scans, inertial measurement unit (
IMU
) data,
GPS
location information, and more. The purpose of automotive data collection is to provide real-time, accurate vehicle and environmental data as input to the autonomous driving system for perception, decision-making and control.
2. Car calibration:
Car calibration is to adjust and optimize the parameters and configuration of the vehicle so that it has good performance and safety in different environments and conditions. Calibration mainly includes calibration requirements, calibration plan formulation, calibration development and verification, extreme environment development and verification, calibration review, calibration release and other steps. In vehicle calibration, the collected vehicle data is used to evaluate and adjust the vehicle's system perception, decision-making and control algorithms to adapt to changes in different roads, weather, traffic conditions, etc. Calibrated parameters can include sensor calibration, vehicle dynamics parameters, obstacle detection and tracking algorithms, etc.
For example: engine management system calibration (including adjustment of ignition timing, fuel injection volume, valve timing and other parameters to achieve optimal power output, fuel economy and emission control), transmission system calibration (including automatic transmission shift logic , clutch operating point, transmission ratio selection and other parameters adjustment to provide smooth shifting and driving comfort), brake system calibration (including adjustment of the material and geometric parameters of brake discs and brake pads to achieve the best Braking performance and durability), suspension system calibration (including adjustment of suspension hardness, damping characteristics and other parameters to provide good suspension comfort and handling stability), etc.
3. Data reinjection: Data reinjection refers to injecting the collected actual vehicle data into the vehicle controller or domain control for simulation and testing. For example, through data refeeding, actual driving scenes and situations can be reproduced, and the performance and algorithms of the autonomous driving system can be evaluated and verified. The data fed back can include sensor data, vehicle status information, driving behavior data, etc. The purpose of data refeeding is to conduct large-scale testing, verification and optimization in a simulation environment to accelerate the development and deployment process of autonomous driving systems.
These steps play a key role in the development and testing of the entire vehicle. Through data collection, calibration and feedback, the overall performance, safety and adaptability of the vehicle can be improved, and its commercialization process can be accelerated.
02.
Market background and demand
As of
2021
, the automotive data collection market is already quite large and is expected to continue to grow in the coming years. According to data from market research companies,
the global automotive data collection market size in 2021 will be approximately
US
$
15
billion, including related products and services such as in-vehicle sensors, smart vehicle platforms and data analysis solutions. This is mainly due to the promotion of digital transformation in the automotive industry and the increasing emphasis on vehicle data.
The growth of the automotive data collection market is driven by several factors. First, automakers and technology companies are placing increasing emphasis on the collection and analysis of vehicle data to provide a better driving experience, vehicle safety and vehicle maintenance. Secondly, the rise of new energy vehicles and autonomous driving technology has created greater demand for data collection and analysis. Additionally, increased connectivity of vehicles with other smart devices and systems is also driving the growth of the automotive data collection market.
The development of new energy vehicles and autonomous driving technology has a significant impact on vehicle data collection, which is mainly reflected in the following aspects:
1. Increase in the amount of data:
The introduction of new energy vehicles and autonomous vehicles has greatly increased the amount of data generated by vehicles. Electric vehicles contain a large amount of data related to batteries, motors, energy management, etc., while autonomous vehicles collect information about the surrounding environment through various sensors (such as lidar, cameras, and radar). The large-scale data generated by these vehicles need to be collected, transmitted, stored and analyzed to support vehicle control, performance optimization and safety requirements.
2. Diverse data types:
New energy vehicles and autonomous driving technology have introduced various new data types. For example, electric vehicles need to collect battery status and performance data for battery management and charging strategy optimization. Autonomous driving vehicles need to collect environmental perception data, vehicle status data, behavioral data, etc. to support autonomous decision-making and driving control. Therefore, automotive data collection systems need to be able to process and analyze diverse data types to provide more comprehensive information support.
3. Data security and privacy protection:
With the increase in the amount of car data, data security and privacy protection have become more important issues. New energy vehicles and autonomous vehicles carry a large amount of sensitive information, such as vehicle location, driving behavior and vehicle performance. Therefore, the automotive data collection system needs to take corresponding security measures to protect the confidentiality and integrity of the data and prevent the data from being illegally accessed and abused.
4. Increased real-time requirements:
Autonomous vehicles have more urgent needs for real-time data. For example, vehicles need to collect and process sensor data in real time in order to accurately perceive and make decisions about the surrounding environment. Therefore, the automotive data collection system needs to have fast and efficient data collection and processing capabilities to meet real-time requirements.
5. Safety and driving assistance systems:
Calibration of autonomous vehicles requires attention to parameter adjustment of safety and driving assistance systems. This includes the calibration of automatic braking systems, adaptive cruise control, lane keeping assist and other functions to ensure their safety and reliability in different road and traffic situations.
In general, the development of new energy vehicles and autonomous driving technology has promoted the progress and application of automotive data collection technology. They have put forward higher requirements on data volume, data type, data security and real-time performance, and promoted the innovation and development of data collection technology. At the same time, it also provides a richer foundation for areas such as intelligent transportation, vehicle management, and driving assistance.
03.
Market status and industry pain points
With the development of new energy vehicles and autonomous driving technology, the current automotive data collection industry is facing some new pain points and challenges, including the following:
1. Data standardization and interoperability:
Since different automobile manufacturers and technology suppliers adopt different data formats and interface standards, interoperability issues in data collection have resulted. This complicates data collection and integration, limiting the effective use and sharing of data. The industry needs broader data standardization efforts to promote data interoperability and interoperability.
2.数据安全和隐私保护：
随着车辆数据的增加和敏感性的提高，数据安全和隐私保护成为行业关注的焦点。汽车数据采集系统需要采取严格的安全措施，以保护数据免受恶意攻击和非法访问。同时，需要确保车主和驾驶员的个人隐私得到妥善保护，符合相关法规和政策要求。
3.数据质量和准确性：
汽车数据采集过程中可能面临数据质量和准确性的挑战。传感器的误差、噪声和故障可能导致数据不准确或不完整。数据采集系统需要具备高精度和可靠性，以确保采集到的数据质量达到要求。
4.处理和存储的挑战：
随着数据量的增加，对数据的处理和存储也带来了挑战。大规模的数据处理需要高性能的计算和存储资源，以支持实时分析和决策。此外，对于长期存储和管理大量数据也需要解决数据存储和备份的问题。
5.法规和合规要求：
汽车数据采集涉及到一系列法规和合规要求，包括数据隐私、数据安全、数据使用和共享等方面。行业需要遵守相关法规和政策，并确保数据采集和处理过程符合法律和道德的要求。
解决这些痛点和挑战需要行业各方的合作和努力，包括制定更统一的数据标准、加强数据安全和隐私保护、改进数据质量和准确性、提供高效的数据处理和存储解决方案，以及遵守相关法规和合规要求。
同时，在新能源汽车和自动驾驶的发展过程中，标定也同样面临一些困难，包括以下方面，
1.复杂性增加：
新能源汽车和自动驾驶系统的复杂性比传统汽车更高，涉及到多个子系统、传感器和算法的集成。解决方案是建立全面的标定策略和流程，确保各个子系统和组件的相互配合和最佳性能。
2.大规模数据处理：
新能源汽车和自动驾驶系统产生的数据量巨大，包括传感器数据、高精度地图数据、车辆状态数据等。处理这些大规模数据需要强大的计算和存储能力，解决方案是采用高性能的计算平台和云计算技术，以加快数据处理和分析的速度。
3.多样化的工况和环境：
新能源汽车和自动驾驶系统在不同的工况和环境下工作，如不同的天气条件、路面状况、交通情况等。解决方案是进行大量的实地测试和仿真，在各种工况和环境下收集数据，并进行标定和优化，以确保系统在各种条件下的稳定性和可靠性。
4.安全性和可靠性要求：
新能源汽车和自动驾驶系统对安全性和可靠性要求非常高，任何错误或不准确的标定都可能导致严重的后果。解决方案是建立完善的安全测试和验证流程，包括模拟测试、实地测试和验证，以确保系统在各种情况下的安全性和可靠性。
5.快速迭代和更新：
新能源汽车和自动驾驶技术发展迅速，要求标定工作能够快速适应新的技术和算法。解决方案是建立灵活的标定流程和工具链，以便快速调整参数和算法，实现快速迭代和更新。
总体而言，面对新能源汽车和自动驾驶的发展，标定困难的解决方案主要包括建立全面的标定策略和流程、采用高性能计算平台和云计算技术、进行大规模数据处理和分析、进行多样化的实地测试和仿真、建立完善的安全测试和验证流程，以及建立灵活的标定流程和工具链。
回灌的验证过程同样重要，在汽车标定数据回灌的过程中，存在的一些常见的痛点和挑战有：
1.数据质量和一致性：
回灌过程中使用的数据质量和一致性是一个关键问题。确保回灌的数据准确、完整且一致，对于正确的标定和系统优化至关重要。数据采集设备和方法的选择、数据处理和校正的准确性都会对数据质量产生影响。
2.数据处理和存储容量：
汽车系统产生的数据量庞大，需要进行有效的数据处理和存储。处理和分析大量的回灌数据可能需要高性能的计算和存储设备，以确保数据的及时性和准确性。同时，需要考虑数据存储的可靠性和安全性。
3.数据回灌时机和频率：
确定数据回灌的时机和频率是一个重要的问题。过于频繁的回灌可能会对系统的性能产生负面影响，而过于稀疏的回灌可能无法捕捉到系统的动态特性。因此，需要合理选择回灌的时机和频率，以平衡系统性能和数据采集的成本。
4.标定和回灌的自动化：
标定和数据回灌过程通常需要手动操作和人工干预，这可能导致效率低下和错误的风险。实现标定和回灌的自动化是一个挑战，需要开发相应的算法和工具，以减少人工操作和提高效率。
04.
数据采集、标定与回灌优秀解决方案与产品介绍
前文已充分介绍了随着新能源汽车和自动驾驶技术的发展，当前汽车数据采集、标定与回灌行业面临一些新的痛点和挑战，下面着重介绍下针对这些困难而涌现出的解决方案与产品：
1.标定软件的产生：
随着标定工作的日趋复杂，为提高汽车开发效率，有许多专门设计的标定软件可用于新能源汽车和自动驾驶系统的参数调整和优化。这些软件提供直观的用户界面，允许工程师轻松配置和调整系统参数，同时提供实时数据监测和可视化，以帮助分析和评估标定效果，诸如
ETAS
公司的
INCA
、
INCA FLOW
与
VECTOR
的
CANape
等，下面以
INCA
为例对当前先进的标定工具进行展开介绍。
ETAS INCA：
ETAS INCA
是
ETAS
公司旗下一款汽车标定测量诊断工具，支持各种汽车控制单元的参数调整和优化，具备强大的功能与优秀的易用性、兼容性，能够极大地提高标定效率，被广泛应用在各主机厂、
Tier1
的
ECU
项目开发过程中。目前全球有超过
5
万名用户在日常工作中使用。
INCA
不仅可以应用在传统的发动机电控开发领域，也可以应用在新能源整车、电机电池控制和处理器标定领域，并且在新的版本中增加了专门针对新能源车的功能。
INCA
基础软件和各种插件
INCA性能特征
￭集成数据监控
/
记录及标定更改等功能于一体，是集成化的标定开发工具；
￭支持
CAN
通讯及
CCP
协议；
￭可灵活定义所需监控及记录的参数；
￭可灵活制作工作界面支持
Copy&paste
功能；
￭支持很多专用数据采集设备，如
ES590
等，使用灵活；
￭标定文件的比较
/
切换等功能，输入输出不同格式的标定数据等功能；
￭标定参数的查找
/
更改
/
图形化功能；
￭可按标定功能定制标定参数模块；
￭可整体输入
/
输出不同配置的工程文件，利于团体协作；
￭数据回放及数据后处理功能；
INCA界面展示及名词简介
￭
DBC
为
CAN
的数据库文件；
￭
Experiment
文件是各类数据的记录
/
显示配置，是一个预先设置好的窗口，里边包含为实现测量和标定任务所需要的变量和匹配值；
￭
Project,
即项目包含了所有的匹配值和数据，这些匹配值和数据反映了一定版本的代码和匹配值。项目包含两个文件
.a2l
和
.hex(
或
.s19)
；
￭
A2L
文件，变量地址文件（变量名
+
地址），用来描述测量变量和标定参数；
￭
Hex
文件，包含了由数据和代码组成的
ECU
控制程序，可以直接下载到
ECU
中运行；
￭硬件配置，可以选择硬件设备及硬件参数
2.数据采集系统：
数据采集系统用于收集车辆和环境数据，以支持标定过程。这些系统通常包括传感器、数据记录设备和相关软件。工程师可以使用这些系统采集车辆传感器数据、高精度地图数据等，并将其用于标定和验证过程中的仿真和实验室测试。为解决待采信号种类多、数据量大、占用算力等问题，数据采集系统也进行了升级，下面对当前使用较广泛的
ETAS
数采产品做简要介绍。
ETK/FETK/XETK
ETK
产品家族是一系列安装在
ECU
里面或外面的小型的、可靠的、高性能的硬件设备，使得软件开发和标定工程师可以直接访问
ECU
来控制变量和参数。
使用ETK的优势:
￭开发软件与产品软件结合紧密，可与标定软件
INCA
等协同使用，提高标定效率
￭
ETK
具有非常高的运行时效性
￭一个接口可以在整个开发周期中使用
￭
ETK
速度和数据吞吐量远超
XCP
，可解决海量数据采集传输速率跟不上及占用
ECU
算力等问题
￭
ETK
帮助工程师减少开发和标定的工作时间和工作量
ECU和总线接口模块：
多信号多系列可解决待采设备多样化、通讯信号多样化与通讯协议多样化的问题，提高数据采集效率。
3.数据回灌系统：
It is also important for development to process the collected actual vehicle data through algorithms and then inject it into the autonomous driving system for simulation and testing to verify the algorithm or reproduce some scenes and states: through data reinfusion, actual driving can be reproduced Scenarios and situations, and evaluate and validate the performance and algorithms of autonomous driving systems. The re-injected data can include sensor data, vehicle status information, driving behavior data, etc. The amount of data is large and complex, and the re-injection verification is more difficult than the collection. Therefore, powerful data feedback software is an indispensable and important tool in the development process.
Next, take
ETAS
's
ADAS
data refeeding solution as an example to explain the actual solution for data refeeding.
The picture above shows
ETAS
's high-performance
PC-
based
recirculation solution, which has the characteristics of scalability and high bandwidth. In order to cope with
the increasing number of
ADAS-
related sensors, the recirculation system can support the input of up to
16
cameras,
12
radars and
4
lidars, as well as the synchronous acquisition and multi-bus interfaces (Ethernet,
CAN
,
FlexRay
). Feedback, based on recorded data, plays back defined scenarios to perform batch testing. This solution can use the data collected by the vehicle and feed it back to
the ADAS ECU
to verify the
ADAS
algorithm
that is modified and iterated one after another
.
The ETAS
data refeeding solution solves the problem of collecting hundreds or thousands of vehicle mileage for
ADAS
software release and regression testing
in actual development
, and improves the execution release and regression of virtual test systems and hardware open-loop (
HoL
) systems. Testing efficiency also solves the problem of very time-consuming and expensive testing of specific edge cases on vehicles.
05.
The future of data acquisition and calibration
Data acquisition, calibration and data refeeding play an increasingly important role in the automotive industry and will continue to grow and evolve in the future.
Here are some future prospects for data acquisition, calibration and data refeeding:
1. Diversification of data collection:
With the development of automotive technology, the diversity of data collection will become an important trend. In addition to traditional vehicle sensor data, more external data sources will be involved, such as Internet of Vehicles data, high-precision map data, traffic data, etc. This will provide a richer and more comprehensive data source for calibration work to better optimize and adjust vehicle performance and behavior, and can even be used to optimize traffic flow and solve traffic congestion.
2. Automated calibration:
As autonomous driving technology matures, automated calibration will become an important development direction. Automated calibration uses technologies such as machine learning and artificial intelligence to automatically adjust and optimize vehicle parameters through the analysis of large amounts of data and model training. This will greatly reduce manual intervention and time costs, and improve the accuracy and efficiency of calibration.
3. Real-time calibration and online updates:
As vehicles become more and more intelligent and connected, real-time calibration and online updates will become more important. Vehicles can obtain the latest calibration parameters and optimization strategies in real time by connecting to the cloud system to adapt to different driving environments and road conditions. This will enable vehicles to be more adaptable and performant, and able to quickly respond to changing needs and challenges.
4. Data privacy and security:
As data collection increases, data privacy and security issues will become more important. The data collected may involve users' private information, so corresponding security measures need to be taken to ensure the safe transmission, storage and use of data. At the same time, data security during the calibration process also needs to be guaranteed to prevent data from being tampered with or used maliciously.
5. Calibration standards and specifications:
With the popularity of data collection and calibration, the formulation of calibration standards and specifications will become an important development direction. Developing unified calibration standards and specifications will help ensure the consistency and comparability of calibration and improve the efficiency and quality level of the entire industry. This will promote data sharing and cooperation and drive further development of calibration technology.
6. Intelligent and adaptive data refeeding:
Data refeeding will become an important means to achieve intelligence and adaptability. By feeding actual operating data back into the vehicle system, the system can continuously learn and optimize, adapt to different driving environments and driver behaviors, and provide a more intelligent and personalized driving experience.
7. Reinjection realizes personalized driving experience:
Through data refeeding, the vehicle can be optimized according to the driver's personalized preferences and habits. For example, based on the driver's driving style and preferences, the vehicle can adjust various settings such as seats and audio systems to provide a more comfortable and personalized driving experience.
8. Refeedback enables system optimization and fault diagnosis:
Through data refeedback, vehicle manufacturers and technology providers can analyze and evaluate the performance of vehicle systems and discover potential problems and faults. This will help improve system design and development and provide more effective fault diagnosis and maintenance strategies, improving vehicle reliability and maintainability.
-end-
Author of this article: Wang Yingfei
, autonomous driving data collection product manager
This column is
a neutral technology popularization column created
by
Automotive Electronics and Software
. It will systematically explain the key challenges and engineering practices under software-defined cars.
Welcome to subscribe to this column!
京公网安备 11010802033920号