An Airbag ECU Contradicts Itself: Discrepancies and Reverse Engineering of EDR Data

Abstract

The retrieval and interpretation of EDR data has evolved into an important area of accident reconstruction over the last few years.

These data are considered to be a reliable source of circumstances of an accident. However, in a current case, EDR data show discrepancies in velocity, steering wheel angle and acceleration values.

The accident occured on a highway, where a Ford Fiesta was hit by a fast approaching Audi A4. The essential question was whether the Fiesta performed a lane change prior to the collision. Unfortunately, the Audi was not equipped with an EDR which could be read out with any tools currently available. However, DTC entries could be used to estimate the collision speed. The EDR data of the Ford Fiesta could be read out successfully. In order to investigate the assumed lane change, the steering wheel angle during the 5 seconds recorded prior to the collision was examined. These values can be used in relation to velocity and lateral acceleration values to check the overall data integrity. Unexpectedly, the EDR report showed steering wheel values that were incompatible with the velocity and lateral acceleration data. It was noticable that the airbag ECU contradicted itself. Some of the EDR data, either the steering wheel angle or the acceleration values recorded, had to be incorrect.

For further analysis, the airbag system with CAN bus and satellite sensors were examined to show the data flow from various sensors to the airbag ECU. For that purpose, the corresponding CAN messages were decoded and driving tests were performed to gain a deeper insight into the offset compensation algorithms involved. The conclusion was that, in some cases, sensor data can be inaccurate and can lead to false interpretations.

The contribution illustrates the reverse engineering of the sensor data collection process within the vehicle and offers conclucions that might be useful to avoid false interpretation of EDR data.

 

More information

Main author

Ingo Holtkötter

Co-Authors

-

Type of media

PDF

Publication type

Lecture

Publication year

2023

Publisher

EVU

Citation

-