This article discusses ADAS testing and explains how ADAS systems and autonomous vehicles are tested. We'll cover the topic in-depth enough for you to:
- Understand why it is important to test ADAS systems.
- Learn how ADAS systems and autonomous vehicles are tested.
- See what type of equipment we use for ADAS testing.
The ADAS testing imperative
The complex technologies that underpin ADAS need to be tested, and not just in software models, but in the field, with vehicles and human drivers interacting on real roads under realistic conditions. For this, high-end data acquisition hardware and software must be installed inside these vehicles.
Automotive testing engineers have been instrumenting vehicles for nearly a century. The current challenge is that they need to test interactions between multiple vehicles (and pedestrians) in real time. When cars are moving, it is impossible to pass wires between them. The vehicles and the ground station must be linked together wirelessly, and the connection must be fast and robust.
In addition, data from multiple test vehicles moving in three-dimensional space must be precisely aligned in time, so that the data can be analyzed meaningfully. Time precision at the microsecond level is the benchmark. We also need to know the exact distances and attitudes between cars and important objects around them.
For vehicles to one day be capable of driving 100% autonomously in all conditions, they must be able to detect and adapt to a wide variety of conditions:
- expected and unexpected movements of other vehicles
- pedestrians on the traffic lane and crosswalks
- debris and unexpected materials on the traffic lane
- undocumented temporary changes due to construction or pavement modifications.
- And many more
It's a long list with almost incalculable variations. Simulation plays an important role in testing, but testing in real conditions is essential. This is why new testing methods have been developed in the world of driver assistance systems, as described in the following sections.
Developing ADAS systems requires complex testing, including the ability to control and calculate relative positions between multiple vehicles and objects in real time, on a large test track. Real-world interactions between multiple vehicles must be tested and analyzed under a staggering number of possible conditions.
In fact, it's not just about acquiring data. A wide range of systems are required to perform ADAS testing. Complex tests involving multiple vehicles and objects require modern test tracks and equipment such as:
- driving robots
- GPS/GNSS
- IMU and INS sensors
- high-speed car-to-car and car-to-base wireless data networks
- ADAS targets
- test tracks
Completely new ADAS test tools have been developed and are available on current test tracks. There are also partially and fully standalone testing tools, primarily in the world of mock testing. Mock testing is essential, but in this article we will focus on real-world testing.
In the second part of the ADAS article series, we described the range of sensors installed on vehicles equipped with ADAS: cameras, RADAR, LiDAR, SONAR, IMU and GNSS navigation systems. To test them, even more precise versions of these same sensors must be installed on the test vehicles. We will cover them all in this article.
Driving and steering robots
Driving robots are used for the computerized operation of motor vehicles. They are typically designed so that they can be installed on a standard car, truck or bus without the need to remove the steering wheel or significantly modify the vehicle.
They are often attached to the vehicle windshield using very powerful suction cups. They do not interfere with the standard steering wheel airbag. Most of these robots offer optional or standard brake, accelerator and clutch pedal controls. They have digital interfaces with a variety of navigation and automation systems on the market.
Driving robots offer a level of repeatability that human drivers simply aren't capable of achieving. They can perform the exact same maneuvers every time. This means that tests are carried out according to standards and test results can be more easily compared and analyzed.
Steering robots are essential for standard NHTSA rollover tests, such as the Fishhook and J-Turn, UN Reg 13-H/FMVSS 126 fall testing, and more.
There are also braking and pedaling robots, separate from driving robots. Most manufacturers in this area make both. Companies involved in driving robot technologies include AB Dynamics, Humanetics, and Stähle.
Inertial Measurement Units (IMU) and Inertial Navigation Systems (INS)
The terms IMU and INS are often used interchangeably, but they are actually two parts of the same thing. The IMU is an actual sensor, made up of accelerometers, gyroscopes and often magnetometers. These outputs are passed to the INS, which uses them to calculate linear velocity, linear position, angles, etc.
Inertial measurement units are used to measure and emit several parameters such as:
- acceleration
- The direction
- Angles
- gravitational forces.
They are generally composed of three accelerometers and gyroscopes: one for each axis, defined as roll, pitch and yaw. They can also hold three magnetometers. High-quality IMUs provide precise measurement of all vehicle dynamics, including slip angle. All data is transferred to a master system which obtains measurement results in real time during the test.
There are several core technologies used by the different IMUs on the market today:
- FOG – Fiber Optic Gyroscope
- RLG – Ring Laser Gyroscope
- MEMS - Micro-electro-mechanical systems
Each of these basic technologies has advantages and disadvantages. RLG and FOG sensors have long been the most efficient, but they come at a price. MEMS sensors are significantly cheaper but still offer very good performance.
Adding IMUs to a GPS/GNSS system adds blind fixing capability since the IMU provides attitude, speed and position information even when GPS satellites are out of sight, e.g. , when crossing tunnels or under bridges.
Dewesoft's DS-IMU1 and DS-IMU2 combine gyroscopes, accelerometers, magnetometers, pressure sensors and a high-speed GNSS receiver. Coupled with sophisticated algorithms, they provide highly accurate and reliable navigation and orientation at an output data rate of up to 500 Hz. The DS-IMU2, in particular, is compatible with all of these types of tests:
- ADAS
- Braking/Acceleration
- Vehicle dynamics
- Lane change
- Control loop
- Chassis development
- Comfort tests
- Noise passing
- FuSi (functional safety)
Companies that manufacture IMUs for ADAS applications include Dewesoft, Genesys, OxTS and Xsens.
GPS/GNSS sensors
High-precision GNSS/GPS positioning systems are an absolutely essential part of today's ADAS testing. These systems are used to measure the relative positions and speed of vehicles and objects with high accuracy and rapid refresh rates.
One or more vehicles can be monitored in three-dimensional space. Synchronized data from all vehicles provides very precise information on the position and distance between them and/or to a fixed object.
Improve accuracy with Real-time Kinematics (RTK)
Real-Time Kinematic is a technique used to increase the accuracy of GNSS positions using a fixed base station, which wirelessly sends correction data to a mobile receiver.
Many GNSS and IMU navigation devices can be upgraded with RTK (Real Time Kinematic) technology. RTK technology improves positioning accuracy up to 2 cm (0.79 in) through the combination of a GNSS receiver and an additional base station.
Using corrections from a fixed base station, the GPS device can fix the antenna position to within 1 or 2 cm. The technique involves measuring the carrier phase of the satellite signal, which is then subjected to sophisticated statistical calculations in order to align the phase of these signals. These fixes eliminate the majority of normal GPS-type errors.
This alignment process goes through three phases:
- the acquisition,
- the “floating” mode of ambiguity, and
- the “Fixed” mode of ambiguity.
The accuracies in "Float" mode are of the order of 0.75-0.2 m and 0.01-0.02 m in "Fixed" mode. The correction signal is normally sent at one second intervals, but can be increased if necessary to reduce the required data rate.
Wireless data transfer using robust WIFI technology
Robust WLAN solutions are used to maintain communication between vehicles and between vehicles and ground stations. There is theoretically no limit to the number of vehicles within this measure - usage is only limited by the aggregated WLAN bandwidth.
Systems like Dewesoft's DS-WIFI are Wi-Fi modems for long-range wireless data transfer between digital data acquisition systems. It is ideally suited for testing moving objects and remote measurement applications.
Compared to wired connections, data acquisition via WiFi offers freedom and flexibility. Wireless connections are absolutely necessary when performing ADAS testing. WiFi antennas are mounted on vehicles using heavy-duty suction cups.
This allows test data to be transferred to the ground station in real time. This also allows ground station personnel to observe the test in real time and even control the test setup remotely.
Wireless transfers can reach a range of 2 km (1.24 miles) in line of sight. Selectable dual-band frequencies of 2.4 GHz and 5 GHz are supported. As the tests are carried out outdoors in a wide variety of environments, the antenna and housings are integrated into robust, waterproof, industrial-grade metal housings.
ADAS targets
ADAS vehicle soft targets, or “soft targets,” are designed to resemble vehicles and other objects on the road, including cyclists and pedestrians. They are realistic enough to convince the camera and other sensors in the test vehicles that they are real. But colliding with them does not destroy them or damage the expensive test vehicles.
Guided soft targets
These targets are remote-controlled vehicles made of soft materials that break easily in the event of a collision. They look like a car or truck but are empty inside. After a collision, they can be rebuilt in minutes and ready for the next test.
Some models, like those made by Humanetics, are made up of several inflatable parts. These 3D “dummy” vehicles typically include a fabric “skin” that makes them look very realistic to the vision systems of the car being tested. Some soft targets include remotely controllable headlights, brake lights and turn signals, allowing a wide range of ADAS tests to be performed in a safe environment.
Mounted on a very low propulsion platform, which can be driven on without damage, the guided soft targets can be driven at up to 80 km/h (~50 MPH) and include braking controls. Some manufacturers specify that they can be checked to within 10 cm. ADAS soft targets are very important for Euro NCAP testing as well as a wide variety of ADAS tests in general.
Soft targets for pedestrians and cyclists
Available in adult and child sizes, these targets are remote-controlled human-shaped mannequins. Some of these human targets are also articulated, meaning their legs move like humans as they move down the test track road.
These targets can be mounted on a very low propulsion platform, or on an arm connected to the propulsion platform, so that test cars hitting them are not damaged by passing over the platform. form of propulsion. Mannequins can be configured as pedestrians, cyclists, scooters, etc. Some models are propelled on a remote-controlled platform, while others are towed using flat cables connected to motors located on the side of the track.
Video cameras
Cameras are an indispensable sensor in a wide variety of automotive testing applications, including ADAS testing. Video data from these cameras, in sync with the data, provides important context, meaning and value.
The most critical aspect of adding cameras to the test is that the resulting video data is synchronized with GPS, CAN and analog data. Dewesoft's DS-CAM camera series features a synchronization interface that, combined with DewesoftX software, timestamps each video frame in the data file. Video data is stored with the rest of the data and is displayed and replayed together seamlessly.
LiDAR sensors
LiDAR (Light Detection and Ranging) systems are used to detect objects and map their distances in real time. When a LiDAR sensor rotates, it emits laser beams in all directions. The time it takes for the beams to bounce is measured and used to construct a high-resolution three-dimensional model of its surroundings.
LiDAR sensors are integrated into driver assistance systems and autonomous vehicles because they are essential sensors. But the LiDAR sensors discussed in this section are the most expensive and high-performance ones that are used for ADAS testing. They are typically mounted on the hood or roof of ADAS test vehicles and used in field testing.
California-based Velodyne manufactures cutting-edge LiDAR sensors for ADAS testing and other applications. Data acquisition systems manufacturer Dewesoft provides an interface to capture and synchronize Velodyne's output with the rest of the data. With DewesoftX software, engineers have a complete view of data from analog sensors, CAN bus, video cameras, GPS, LiDAR, and more.
LiDAR systems can be synchronized with GNSS (Global Navigation Satellite System) and GPS (Global Positioning System) devices using the PPS (pulse per second) signal. In this configuration, the GNSS receiver is the source of NMEA (National Marine Electronics Association) GNSS messages for the LiDAR sensor. LiDAR and GPS/GNSS are connected directly.
Proving grounds for ADAS testing
A proving ground is an area where a vehicle's performance is put to the test. Testing grounds typically span large areas and roads (usually several kilometers or miles) and have garages and related facilities to evaluate the operation of various vehicle systems and parts. Proving grounds, also called "test tracks", have been testing vehicles on and off the road for decades.
Until the 1920s, automobile testing was carried out on city streets and country roads. Proving Grounds move vehicle testing from public roads to controlled, secure and safe testing environments, while simulating a wide range of road types and events, all reflecting or relating to vehicle use. vehicle by customers.
For cold weather testing, the best-known facility is in Arjeplog, Sweden, just 50 km from the Arctic Circle. Ordinary roads and bridges are used in combination with the frozen lake, on which roads are laid in winter, for a wide range of tests. Most of the world's testing grounds are private facilities to ensure safety and keep new designs away from competition.
ADAS software
ADAS software is required to record data from all vehicles and other objects included in the ADAS test. The ADAS software must be capable of acquiring data synchronously from multiple vehicles at the same time and from different types of ADAS sensors. He must also be able to carry out measurements and various calculations such as:
- measure the speed, acceleration, deceleration and heading of an object.
- measure and calculate distances between moving and static objects in real time
- measure and calculate angles between objects
The DewesoftX software, which is included in Data Acquisition Systems and Inertial Measurement Units, offers an add-on called Polygon, developed specifically for the needs of ADAS testing.
Polygon is a 3D virtual platform for testing involving moving and static objects. It was designed especially for ADAS testing and vehicle dynamic testing, which increases safety in traffic. The Polygon add-on provides a visual representation of measurements in three-dimensional virtual space and easy-to-use tools for geometric measurements between multiple static or moving objects. Thanks to its flexibility, it is not only used for testing autonomous automotive vehicles, but also in the marine and heavy machinery sectors.
The Polygon add-on allows you to define multiple object types and perform different calculations between them. The following object types can be defined in the polygon:
- Vehicles
- Simple objects (cone)
- Line
- Multipoint line - route
- Circle
- Movement radius
- Import
- Static objects
- Moving objects
- Mobiles with other objects
- Frozen/thawed on trigger
Once the objects are defined, the Polygon will be able to perform multiple calculations between the defined objects in real time and visualize them appropriately on the screen. These calculations include:
- Distance: calculates the distance between two objects. Both objects may be moving or one may be stationary. All object types are supported for distance calculation.
- Distance X and Y: calculates the longitudinal and lateral distances, starting from the first object. Distances can be calculated between vehicles and simple objects (fixed or moving). If the X and Y position of an object on the fixed coordinate system is required, you can place a simple object in the center (at position x=0, y=0). The polygon then calculates the distances X and Y between this object and the moving object.
- Angle: calculates the heading difference between two objects. It can be calculated between the vehicle and a line, road, circle, travel radius, or another vehicle (the vehicle must always be the second object).
- Gate Trigger: Changes its value from zero to one and returns to zero when a moving object crosses a defined line. The first object should always be a line (representing the "door"), and the second object should be a vehicle or simple object (i.e. a custom point of interest).
- Time: Gives the relative time since the previous time reset in seconds and resets the timer. One of the measurement objects must be a line. The output is changed when the other specified object (vehicle, simple object) crosses the center of the line. A line with a time setting can be used to record passing times on a looped track.
- Time reset: resets the relative timer and outputs the absolute time since the start of the measurement in seconds. One of the measurement objects must be a line. The output is changed when the other specified object (vehicle, simple object) crosses the center of the line. Normally this parameter is used on the start lines of non-looped tracks (drag races, braking tests).
- Radius: Provides the value of the specified movement radius in meters. It uses only the movement radius specified for the measurement object. It can be configured to produce a ray or a reverse ray. An inverse radius is used when we want to avoid large radius values when the trajectory of the moving object is close to a straight line.
ADAS software such as DewesoftX and the Polygon add-on provide an excellent virtual platform for testing all kinds of ADAS scenarios on proving grounds or in real-world conditions. Multiple vehicles can be equipped with Dewesoft data acquisition instruments and displayed in real time using DewesoftX software.
What about ADAS calibration at the consumer level?
In previous sections, we saw how engineers test ADAS systems during development, using real-world testing and software simulations. All of this happens before the car designs are finalized, manufactured, and then delivered to consumers.
But what happens once a new vehicle equipped with driver assistance systems is parked in front of your house? Do ADAS systems need to be checked and calibrated from time to time? As with any other part of the car, the answer is yes.
As you learned in the first part of this article, ADAS systems rely heavily on sensors, which must perform optimally to allow the ADAS system to perform as intended. A powerful processor is also behind the scenes. It processes gigabytes of sensor data and uses it to issue driver assistance warnings (passive safety measures) and take action to keep the vehicle in its lane, and automatically brake and steer the vehicle to avoid a collision (active safety measures), and more.
In fact, automakers are increasingly recommending, and even requiring, that consumers have their car's ADAS system recalibrated in these circumstances.
ADAS warning lights flashing or indicating a fault
This point is obvious: if your car tells you that something is wrong with the ADAS system, it should be checked and brought back to original standards immediately.
Windshield repair or replacement
Several ADAS sensors are mounted near or behind the windshield, so replacing or deteriorating a windshield can affect these sensors. This is why several companies that carry out windshield replacement and repair have embarked on the recalibration of ADAS sensors. American company Safelite Autoglass offers ADAS camera recalibration as part of its windshield repair services. It offers both static recalibrations, where the car is parked and facing a calibrated target, and active recalibrations, where the car is driven on a well-marked road.
Safelite is present throughout the United States. There is probably a company in your city that calibrates advanced driver assistance systems. More and more car dealerships can perform calibration of advanced driver assistance systems to ensure the systems meet the manufacturer's OEM specifications. Today, other companies are also entering the market.
Accidents
A collision causes very obvious damage to the car body. But the impact can also affect systems you can't see, including sensors and the ADAS system. Checking and, if necessary, recalibrating the ADAS system should be part of any collision repair.
Changes in car height
ADAS systems such as parking sensors and others are calibrated for a specific ride height. But if different diameter tires are installed, or if significant changes are made to the car's suspension, the ride height may be changed. It is important that the sensors are calibrated to the actual ride height of the car so that they can perform as intended.
Consumers should always verify that the repair center is trained to calibrate the ADAS system according to the automaker's factory specifications.
What about cybersecurity?
The more connected something is, the greater the risk of hacking. We've seen bad actors hijack the computer systems of banks, cities, and pipeline operators and hold them hostage until a ransom is paid. In a future where most or all vehicles on the roads are interconnected, how can we be sure that a single car or truck...or fleet of vehicles...won't be attacked?
An independent study commissioned by SAE and Synopsis reveals that 30% of automotive manufacturers and suppliers do not have a product cybersecurity program or team. And 84% of respondents are concerned that cybersecurity practices are not keeping pace with the ever-changing security landscape.
Today's cars, trucks, and even motorcycles are essentially "autonomous rolling computer networks" that integrate control systems, entertainment systems, and wireless communications via numerous protocols. Driver and passenger cell phones, tablets and other devices can, and often do, connect to the vehicle for entertainment, communication and navigation purposes. They also expose the vehicle to the wider internet and, increasingly, vehicles have their own wireless link to the internet. The potential for piracy has never been greater.
One of the conclusions of this report is that the technologies that pose the greatest risk are RF technologies, telematics and autonomous driving vehicles.
To combat this risk, SAE developed J3061, the world's first automotive cybersecurity standard. Basically, the JAE J3061 standard emphasizes that cybersecurity should be integrated into vehicle design from the beginning, not added later. The pressure to move new vehicles from design to market as quickly and economically as possible has a powerful effect on OEMs and suppliers. As vehicles become increasingly interconnected, rigorous testing against vulnerabilities at all stages of product design and development must keep pace. Development testing at this stage is crucial.
A review of the ISO 26262 automotive functional safety standard is underway, and a working group consisting of ISO 26262 and SAE J3061 has begun developing a comprehensive cybersecurity standard for tomorrow's vehicles. A United Nations working group on cybersecurity is also looking into this increasingly important topic.
Summary
Each machine must be tested to ensure it operates according to its design and safety protocols. Cars, trucks, and buses are some of the most complex machines the average person will ever operate.
ADAS technology and its safety features were developed to improve the safety of our roads by preventing human errors which are responsible for the vast majority of accidents. Therefore, nothing is more important than ensuring that the ADAS technology itself is safe and reliable. Testing has never been more important.
No comments