On January 23, 2025, the final review meeting for the BMBF Mannheim project EMDRIVE took place in Potsdam.
We demonstrated the prototype implementation of our diagnostic unit, which enables runtime diagnostics for identifying or isolating sporadic software errors on ECUs (Electronic Control Units) in an Ethernet-based vehicle network. The diagnostic unit is implemented as a hybrid hardware/software solution on a Zynq UltraScale+ MPSoC ZCU102 platform. It detects anomalies in Ethernet communication between ECUs and then initiates the analysis of processor traces or reconfigures the tracing configuration. More information on our diagnosis unit can be found on our internal EMDRIVE project page.
For the prototype implementation, a simplified lane-keeping application distributed across three AURIX boards was used as an example. This application receives driving scene input from a CARLA simulation and sends decisions regarding the necessary steering direction back to it. The diagnostic unit leverages the MCDS diagnostic infrastructure and the TAS server from Infineon. Using this setup, deviations can be artificially introduced into the processing, which the diagnostic unit detects.
The accompanying picture shows the LIS team with the setup of the diagnostic unit demonstrator, which was developed by Zafer Attal (2nd from left) with contributions from several students. Dr. Thomas Wild (left) presented the demonstrator at the meeting. Also shown on the right are our project partners from ZF Friedrichshafen, Dr. Stefan Hommes (right) and Dr. Sebastian Lauer (2nd from right), who focused on intrusion detection in in-vehicle networks as part of EMDRIVE.
EMDRIVE is a Mannheim project funded by the BMBF from February 1, 2022, to January 31, 2025. It addresses a platform concept for distributed heterogeneous automotive real-time computing network architectures. One of the specific project objectives was a monitoring and diagnostic concept, for which LIS contributed the aforementioned diagnostic unit.