-
Notifications
You must be signed in to change notification settings - Fork 2
00.3 Development history
The architecture described on page 00.2, developed over time in an iterative nature. In the following, the development stages that our architecture has gone through are shown:
The very first iteration was mainly about rough theoretical planning. A rough path was sketched, which could then be built upon and improved. At this point, the idea was to record CAN data with the help of a WIFI data logger, then store it online and offline (depending on the urgency of the data) on a small computing unit, and subsequently preprocess it. After that, the data would have been processed for visualization and made available via a webapp or microservice on a local server via a WIFI hotspot.
Critical issues resulting from this architecture were:
- Storage capacity
- Storage of non-relational data
- Physical access to SD-Card in datalogger
- Scalability in the future
- Simple visualization
- Ease of use
- High traffic rate
- Local or hosted server necessary
The second iteration still adhered to the use of the WIFI data logger, but also aimed at data acquisition via a local MinIO server. After preprocessing, the data would be stored in a local database (on a laptop, for example). From there, a program could retrieve the data, perform necessary calculations and finally offer them again wirelessly via a microservice or a webapp.
Critical issues resulting from this architecture were:
- Storage capacity on Laptop
- Storage of non-relational data
- Instability of WIFI Datalogger connection
- MinIO server on Windows
- Access to Linux machine in HS network (Timeline)
- Scalability in the future
- Simple visualization
- Ease of use
The third iteration saw the first use of an Nvidia Jetson AGX Xavier board with ROS environment for data logging. After data collection, the sensor data would then be pre-processed and stored in a local database on the AGX along with the camera data. Subsequently, these could have been queried via a WIFI hotspot in the Tableau Desktop visualization tool.
Critical issues resulting from this architecture were:
- Access to Linux machine in HS network (Timeline)
- Complexity and speed
- Ease of use
From all these development steps, the architecture presented on the System Architecture page crystallized as the right approach. Below is a simplified conceptual drawing of the final iteration.