1. Proper nouns
VRU: Vulnerable Road Users, vulnerable road users, including pedestrians, two-wheelers.
RT-Range: True value, used to measure real-time position, distance, speed, direction and other information. For example, the relative distance and speed between two vehicles can be used in the test. RT-Range can be installed on the self-vehicle and the target vehicle.
LaunchPad: LunchPad is a slidable bottom plate with mechanical structure, pulleys, etc., on which a dummy car can be installed, and the speed of the dummy car can be artificially controlled to drag the dummy car forward, and at the same time, its real-time distance, speed, and orientation can be obtained angle and other information.
![](https://img-blog.csdnimg.cn/img_convert/2c3928a79d797cee87f58e6522d24a70.jpeg)
2. Project and team
1. Project background/project introduction?
Currently working on the ADAS project of xx vehicles, the project is mainly divided into two phases, the V1 phase is mainly the testing of sensors and actuators (work content: write test cases, review, release, and then write test plans according to the test cases, including Vehicle restructuring, personnel arrangement, time arrangement, project budget, equipment leasing, etc.).
After that, go to the proving ground to test according to the test cases and plans. After the test is completed, analyze the results and write the test report.
The V2 stage is the verification test of ADAS-related functions (AEB, ACC, LKA, LCA, BSD, DOW, etc.), and this stage also needs to write test cases and release.
The V2 stage is subdivided into V2.1, V2.2, and V2.3 according to the development maturity. Due to different maturity levels, each stage will select some test cases to arrange test planning, test execution, and test report writing.
2. What is the project staff composition and division of responsibilities?
example:
①Products (camera, millimeter wave radar, lidar);
②Development (system development/application software development/underlying software development/domain controller development/hardware development);
③Test (test case, vehicle test, test management) about 30 people: test case group: docking requirements, writing test cases
Vehicle test: real vehicle test, test case execution Test management: equipment/vehicle management, site reservation
④ Product operation center.
3. What is included in the daily work of the test?
Base:
1) Complete the preparation and release of sensor performance test cases according to the functional description and requirements, including Camera, Radar, Lidar, etc.;
2) Complete the preparation of the test plan, including vehicle modification, test plan, personnel arrangement, etc.;
3) Complete the sensor performance test according to the test case;
4) Use the CAN tool to complete the sensor performance test result analysis;
5) Write sensor performance test report and review;
6) Complete the preparation and release of functional test specifications such as AEB/ACC/LKA according to test cases and requirements specifications;
7) According to the development progress, complete the arrangement of each function test plan;
8) Complete the functional verification test according to the test specification;
9) Complete the analysis of functional test results and follow up the progress of problem improvement;
10) Preparation and review of test reports;
11) Responsible for the road test of the automatic driving system, record the problems of the road test, and form a road test report;
12) Analyze road test problems, classify, manage and track test problems;
13) Assist the R&D team in R&D testing and problem reproduction;
14) Complete other tasks arranged by superior leaders, such as: VRU drive system instruction manual, preparation of Lidar operation guide, sensor calibration test, etc.
Advanced:
1) Responsible for building the team test tool framework, writing test auxiliary tools such as test data generation and test log analysis;
2) Responsible for establishing an automated testing framework, and assisting in the completion of automation scripts and use cases;
3) MATLAB/simulink/carsim completes the modeling of unmanned vehicle kinematics and dynamics;
4) Use carsim and simulink joint debugging to realize vehicle queue model communication and configuration;
5) Use control algorithms such as PID/LQR to finally realize the lane keeping assistance function.
4. What is the project cycle?
Example: ADAS usually takes one and a half to more than two years from project approval to delivery.
5. What is the content of the test report?
①Test purpose and test specification;
②Test equipment (including tested vehicle information, data acquisition system, target object, truth system, etc.); for example: vehicle model, sensor version, actuator version, domain controller version, vehicle mileage before and after the test.
③ Test time, place and personnel;
④ Test environment (weather, light, road surface and other information);
⑤Test results (list the analysis of the actual test results of each use case in tabular form).
3. Sensor performance test
1. What is the testing process?
Example: first meet the requirements, and complete the compilation and review of relevant sensor performance test cases according to the requirements.
Then write the test program, including test plan and vehicle modification.
The test plan is mainly to arrange the time plan required for the test according to the test cases; the vehicle modification is to install the sensor to the designated position of the test vehicle according to the demand. The front millimeter-wave radar is generally installed on the front bumper, and the laser radar is generally installed on the car. Top, and then connected to the gateway and the like, it is necessary to read the vehicle chassis signal, body signal and radar signal, and to rent and install the real value system, launchpad, dummy car, etc. at the test site.
The rest is to execute according to the test cases (the functional tests of the V2 stage do not need to install sensors, etc., and the sensors/actuators have been integrated into the test vehicle at this stage).
2. Vehicle restructuring? Real vehicle environment construction?
Items: vehicle preparation, sensor installation specifications, sensor bracket production, sensor wiring harness, power supply (12V/24V/220V), backbone (body/backbone network), Chassis CAN (chassis), gateway forwarding vehicle signals to radar.
3. Which targets can be equipped with RT-Range?
Only the target vehicle is equipped with the real value, and the pedestrian/two-wheeled vehicle uses a dummy/fake bicycle to be installed on the LaunchPAD, which can obtain information such as distance/speed/azimuth.
4. How to measure the performance of the sensor?
Connect the wiring harnesses such as the truth value, radar, and chassis to the car to ensure that the data can be read, then record the data in the CAN tool of the testing machine, and follow the steps of the test case. After all the steps are completed, end the recording and save the data That's it.
5. The recording software?
Radar: CANOE tool, host computer software for radar (if any), and camera for recording scenes (similar to driving recorder).
Lidar: The upper computer software is used to display real-time point cloud images and playback functions, and Wireshark is used for collection.
6. What are the scenarios for sensor performance testing?
![](https://img-blog.csdnimg.cn/img_convert/aae81d41af7725e62cd5e71bfdcb2dca.png)
7. Where are the sensors installed? effective distance?
![](https://img-blog.csdnimg.cn/img_convert/aee9f1a0847a6ec4781ca822dac5670f.png)
8. What are the advantages and disadvantages of each sensor?
![](https://img-blog.csdnimg.cn/img_convert/b3b3887a2fac55203d358dc37d61554b.png)
9. Does the sensor need to be calibrated?
Whether calibration is required before the test depends on which supplier's sensor is used, some need to be calibrated by themselves, and some suppliers have already calibrated.
Example: The calibration of the radar of xx supplier uses its own calibration software, and the process is relatively simple. Just send a string of hexadecimal arrays under the driving state of the vehicle, and it can also be read by sending a hexadecimal array. Whether the calibration is 100%. The calibration of the camera includes internal reference calibration and external reference calibration. There are two main purposes of calibration, one is to eliminate distortion, and the other is to unify the coordinate system. The calibration is to obtain the internal/external parameter matrix. Zhang’s calibration method is used. A photo implementation of N groups of checkerboards. In addition, there are other calibration methods, such as calibration based on natural scenes.
10. How is the coordinate system determined? Where is the origin of coordinates?
In the ADAS test, the right-handed coordinate system is generally adopted uniformly, with the x-axis facing forward, the y-axis facing left, and the z-axis facing upwards. ).
11. What issues does lidar focus on?
①Whether the point cloud quality of high anti-objects such as traffic signs extends up and down and drags back and forth;
②物体边缘是否会出现不明点云;
③高速运动中观察周围路灯杆是否倾斜、卷曲、断裂等情况;
④阳光直射场景噪点是否会增多。
12、数据如何传输(传输方式)?
摄像头:①视频信号转换为CAN信号直接输出;②通过千兆网卡实现通信。毫米波雷达:通过CAN总线给下游发送消息(数据格式为blf、binary log file,一种二进制文件)。
激光雷达:以太网(CAN传输速度最大只有1M/s,激光雷达点云数据量大,因此用以太网)。
CAN总线最大传输速率是1M/s,以太网10M/s,毫米波雷达采集的数据不是很大,采一段数据大概几十兆,所以CAN够用;但激光雷达的点云数据非常大,录一两分钟的数据就有几百兆数据,所以用以太网;LIN速度只有几十kb/s,一般用在:门、方向盘、座椅、空调、照明等。
13、数据分析具体分析哪些方面?报文看哪些?如何分析?
数据分析主要看传感器采集的数据和真值的误差Dx/Dy/Vx/Vy(横纵向距离和横纵向速度),计算平均值、方差。
分析的过程:首先在CANOE工具里导入dbc文件,包括车身dbc,底盘dbc,真值dbc等(dbc就是CAN数据库,里面定义了节点、报文及对应信号的定义及说明等信息),然后将测试采集到的数据(blf格式)通过工具转成csv格式导出,之后在csv文件中筛选出真值和雷达相关数据,主要是横纵向距离及速度,进行比较,计算平均值和方差。
14、数据分析,误差多少是合格的?
例:距离误差在0.7m之内。
15、测试对天气的要求?
天气一般是在晴朗的天气下进行的,测试规范里有对天气的要求:光照强度不低于20000lux,夜间实验光照强度不高于11lux,水平能见度大于1km,可参考C-NCAP2021。此外由于不同传感器在恶劣环境下的表现不同,也会适当涉及雨雾冰雪这种天气。
16、回归测试策略?
根据性能测试结果分析哪些场景下传感器的性能表现较差,挑选典型测试用例进行回归测试。
17、传感器性能如何评价的(评价术语)?
传感器性能测试:横纵向位置测量准确、有效FOV略小、原距离下FOV表现不理想,目标物初次稳定识别的距离散布很大、测试数据没有明显规律、中远距离出现目标短暂丢失现象、对不同目标物的最远探测距离都很稳定、左右两侧误差分布更为集中、横向误差一般且随纵向距离增大而迅速增大不利于目标物的融合、在各个距离段误差一致性更好同时测量的精密度更高,有利于通过补偿缩小系统误差……
最后:
可以到我的个人号:atstudy-js,可以免费领取一份10G软件测试工程师面试宝典文档资料。以及相对应的视频学习教程免费分享!其中包括了有基础知识、Linux必备、Mysql数据库、抓包工具、接口测试工具、测试进阶-Python编程、Web自动化测试、APP自动化测试、接口自动化测试、测试高级持续集成、测试架构开发测试框架、性能测试等。
这些测试资料,对于做【软件测试】的朋友来说应该是最全面最完整的备战仓库,这个仓库也陪伴我走过了最艰难的路程,希望也能帮助到你!