Gwenael Dunand, Intempora S.A副首席技术官
Data logging is one of the key stages in data-driven development in the ADAS/AD domain. Data must first be collected for training, and then data must be collected for validation and testing.
The data logging system, as a part of the in-vehicle measurement infrastructure, is composed of a hardware platform and application software. The hardware must be flexible and performant enough to process and store high-volume data streams from environmental sensors and external devices (see AUTERA, the data logging platform from dSPACE), while the software must be efficient and easy to use.
The logging software can be tailored to a specific hardware or it can be stand-alone, allowing for an easier transition between data logging platforms. This can be especially beneficial when using a low-cost data logger for low-level automation projects and a high-performance data logger from a different supplier in AD applications. If the same software can be kept, significant time is saved on new knowledge acquisition.
What do users expect from the software? Apart from application-specific aspects, there are some basic requirements for in-vehicle data logging software that need to be fulfilled in order for it to work efficiently and reliably.
1. The hardware platform and operating system used need to be supported. This sounds trivial, but especially in the area of embedded platforms, there are many system variations, and often there is no guarantee that new software will be compatible with all relevant processor architectures. An embedded platform might also be changed between two projects, but the software still needs to be used.
2. The sensors, buses, and networks used in the vehicle have to be supported by the data logging software. There is an unbelievably large variety of these devices and protocols in the car, and all of them need to be taken into account by the software. In addition, agile development practices often introduce changes to the interface or to sensors during the project. To not affect the project timeline, the software must be highly flexible (without involving significant additional effort).
3. User action is required during a test drive, so purely laboratory-focused software would not be enough, and there should be easy and quick ways to access the software, especially for key functionality.
4. High efficiency and high data throughput has to be not only guaranteed by the hardware, but also managed and maintained by the software.
5. Parallel, time-correlated data collection across the different sensor streams is crucial. All captured data has to be time-stamped as otherwise it would be nearly impossible to quickly compile the data correctly after the test drive.
Additionally, seamless integration with the rest of the development environment would be beneficial. With all this in mind, is there an appropriate solution?
RTMaps is a complete development and execution tool that is ideal for data logging.
You can put your mind at rest: Yes, these requirements can be met, and the following solution even solves many more challenges than those mentioned above. I am referring to RTMaps: a powerful yet resource-efficient software that can truly accelerate your real-time multi-sensor applications.
RTMaps has an open-data format (.rec), which makes it possible to record data using multiple files. As a result, if you require only a subset of the original complete data streams, just take the appropriate files and replay them inside RTMaps. You can use only the needed data and are not required to use all recorded data every time. This is very useful for saving hard disk space and ethernet bandwidth when sharing data across development teams, partners, and customers.
Not only does RTMaps support its own fully open and documented data format, but it also uses many existing standards, such as MDF4, PCAP, BLF, etc. The table below contains a few of the compatible data logging standards:
|Intempora - RTMaps||.rec|
|CAN, CAN-FD||.blf, .asc, binary|
|SOME/IP, Ethernet data||Binary (.s8), .pcap|
|Video||.avi, .mp4, …|
|Image||.jpg, .png, .tiff, ...|
|GPS traces||.kml, .nmea|
|LIN, FlexRay||On demand|
|ROS 1 and 2||ROSbags with ROSBridge|
Our middleware RTMaps is not only a logging software, but also a complete development and execution tool, which can be used for innovative automotive applications such as autonomous driving developments, robotics, and much more. RTMaps integrates perfectly with the dSPACE product portfolio, including with the AUTERA hardware and the MicroAutoBox III Embedded PC. Despite the complexity of the tasks, RTMaps remains easy to use: its unique graphical user interface allows engineering teams to work on the same application in a collaborative way, sharing several tasks and software functions. This ease of use enables both beginners and experts in software engineering to rapidly develop and control innovative applications that would normally take weeks to code. This is an outstanding advantage, especially in companies that want to boost their competitiveness in new developments in the field of mobility.