FEV smart vehicle fleet heading toward SAE level 5 automated driving

Automated driving: FEV brings technology platform on the road

20. October 2017 | Engineering Service

For OEM, suppliers and developers, automation of vehicles brings with it a wealth of new tasks. At the same time, the classic functions and purchasing criteria fall ever further into the background. As a service provider, FEV supports its clients from the initial concept to start of production in its decision processes associated with these new subjects.
Since 2016, FEV has been bundling all of the steps associated with advanced, fully connected, automated vehicles in its “Smart Vehicle” Center of Excellence. Smart Vehicle includes all sorts of development fields in a rapidly-changing, highly complex environment – from sensor technologies to software algorithms all the way to electrical/electronic architectures and connectivity.
For example, FEV is developing innovative solutions in forward-looking operating strategies, connectivity and cybersecurity in addition to infotainment and driver-vehicle interaction.
The experts have now taken an important demonstration and development vehicle onto the road. Sébastien Christiaens, department manager at FEV, has been keeping track of this project from the Aachen location and he discussed the background with SPECTRUM.

>> Our goal was to develop standardized interfaces and use them to obtain a modular, state-of-the-art development platform

Mr. Christiaens, for a few weeks an automated demonstrator vehicle from FEV has been driving on the roads or on stretches of road for which authorization has been given. What is the background of this demonstration vehicle?

To be accurate, there are three such vehicles driving in the world. The automated smart vehicles are part of an FEV project that we are conducting with our colleagues in the US, Poland, Turkey and our company headquarters in Aachen. Our goal in doing so is to bundle our global expertise and allow others to experience it in a fully automatic development vehicle as a basis for further development, but also for benchmarking. The vehicles show the current status quo that we worked out in recent years in various projects and are now putting together in one vehicle. The smart vehicles thus form the first automatic FEV fleet that we created without any customer orders.

It was possible to take a look at the vehicle driving at this year’s Aachen Colloquium Automobile and Engine Technology. What was to see there?

Right now, the vehicle can safely travel without a driver on a given stretch and the vehicle can suitably react to any events. Thanks to the object recognition implemented, traffic signs are recognized, as are things and people, and appropriate driving maneuvers are started.
At the Aachen Colloquium, we streamed the trip to our exhibition booth. The streaming was done strictly through the vehicle’s network connection. In the process, the vehicle sends data to the cloud. The Aldenhoven Testing Center, where the demonstrator is headed, offers ideal conditions for doing that. At present, an urban testing area is being set up for mobility research. Thanks to the Vodafone-5G Mobility Lab that is also included, this area also has a high-performance network. That is needed to process the large amounts of data.

For example, which sensors are integrated in the vehicle?

We have integrated extensive sensors, including radar, GPS, different types of cameras as well as differential GPS and LIDAR and a vehicle-2-network connection. Thanks to these sensors and interfaces, we are able to perceive the immediate environment of the vehicle as well as to anticipate the oncoming road and traffic conditions on a longer range. Using intentional redundancies in varying technologies, we can thus eliminate the shortcomings of one type of sensor by enriching and comparing the results using those from other systems. This intelligent combination of the different sensor information, referred to as sensor fusion, is a key element in the vehicle environment detection and localization process. Another of our goals was to develop standardized interfaces and use them to obtain a modular, state-of-the-art development platform. Thus, individual sensors can be swapped out with as little effort as possible in later benchmarking activities.

 

What is the basis for issuing driving commands?

The driving commands themselves are managed using a “decision making algorithm” we developed ourselves. This algorithm has three main parts: perception, planning and decision/action.
This algorithm ensures that the vehicle moves safely. In doing so, we draw amongst other on the expertise of our American colleagues, who already successfully completed automation projects in the past. As a modular development platform, the vehicle is actually equipped with two types of powerful embedded controller hardware. This allows us to test and compare different types of control algorithms, for example a rule based approach and a machine learning/artificial intelligence approach.

If sensors collect environmental data, communication between the vehicle and this environment is surely just an additional logistics step. To what extent have you already planned this aspect?

Communication between the vehicle and its environment is mandatory for automated vehicles. Even today, the vehicles already have a Vehicle-to-Everything („V2X“) connection. The intelligent connection unit – abbreviated as iCU – is based on microservice architecture and processes data and information from all sorts of control units and sensors. The FEV iCU is in a position to process data from vehicle-to-vehicle („V2V“) communication via DSRC. Thanks to the microservice architecture, integration of corresponding 5G standards („C-V2X“) will be possible directly, as soon as they are available. Intermediate data aggregation and data conversion services harmonize the data sets and formats, which often differ a lot from each other.

>> With our Cyber Security Gateway, we offer an important tool to prevent cyber attacks

How are you handling cyber- security when you do this?

Cybersecurity in vehicles is actually one of the greatest challenges. As long as the vehicle lacks a vehicle-2-X connection, the number of ports for attack is still relatively easy to look at. The biggest danger comes from the OBD interface but also from the infotainment system. As soon as the vehicle starts working within a network, the sources of danger grow exponentially. With our Cyber Security Gateway, we offer an important tool to prevent cyberattacks. The cybersecurity gateway is linked with the vehicle’s communications bus in order to detect and prevent malicious attacks. It can also be used as a firewall between external interfaces and the vehicle bus. In addition, FEV works with leading world manufacturers to implement so called Hardware Security Module –HSM- and TPM technologies, known as Trusted Platform Modules, for the automotive industry for secure booting and secure over the air (OTA) software updates to name a few applications.

The cybersecurity gateway is linked with the vehicle’s communications bus in order to detect and prevent malicious attacks. It can also be used as a firewall between external interfaces and the vehicle bus.

What steps are now coming up with the smart vehicle fleet?

As a powerful tool to address important challenges such as integration of new functions, interfaces, and components, these vehicles offer a broad range of possibilities for FEV as well as for our partners and customers.
At first, they have an important role in the development and improvement of our control algorithm as they offer a flexible platform for our engineers to make their innovations in ADAS and automated driving features more tangible. This flexible platform can obviously also be offered to our partners and customers as basis for commonly developed “Proof of Concepts” to demonstrate new technologies or features for example.
As already mentioned above, our automated vehicle fleet is heavily involved in our benchmarking activities. Both for sensor benchmarking as well as for system and overall vehicle benchmarking.
For benchmarking, the most important part is reproducibility of results and test runs, and automation is very important for that. In this case, our smart vehicle is not the technology platform for the test vehicle but is acting as the testing tool. Driving maneuvers of the target vehicle can be automated and done reproducibly and it can be verified that the test object operates reliably.
Finally yet importantly, the vehicles are also supporting the overall calibration, testing and validation activities we offer for ADAS and AD development. For example, we are working on integrating these vehicles into our virtual testing tool chain and environment, among other things as vehicle in the loop platform. FEV also uses these vehicles as a means to develop its Big Data tools and services through extensive data collection and analysis
In general, our vehicle fleet will constantly evolve, helping us to accelerate the delivery of innovative solutions for our customers. The fact that we are developing these vehicles worldwide definitely helps to provide a global answer to questions related to largescale deployment of this technology.

Mr. Christiaens, thank you for the interview!

>> In perspective, we will use the smart vehicle fleet not just to do further development of the individual systems, but also as an important testing tool for level 4 and level 5 functions

 

[addtoany]