SensFlow enables reproducible ADAS validation

The validation of modern ADAS and AD functions requires recordings of real-world test scenarios that can be used reproducibly in the lab. SensFlow from IPG Automotive addresses exactly this challenge: The software suite injects recorded sensor and vehicle bus data deterministically into ECUs and thus enables realistic hardware-in-the-loop tests without the need for extensive real-world test driving.

In the face of the growing complexity of advanced driver assistance systems (ADAS) and automated driving (AD) functions, test and validation requirements are increasing. Sensor setups become more heterogeneous, data rates rise and the number of relevant scenarios continues to grow. At the same time, critical situations need to be reproduced with precision to reliably evaluate localization, environment detection and overall functionality.  

Deterministic replay and exact timing 

SensFlow supports deterministic replay of various sensor data streams, from radar and ultrasonic sensors to cameras and lidar, as well as the replay of CAN and Automotive Ethernet communication. A key aspect is the exact timing with reliable timestamp synchronization to ensure that the data enter the ECU exactly as they were recorded during the real-world test drive.

The modular architecture facilitates adaptation to different ECU environments, sensor setups and data formats. In this way, SensFlow can be utilized flexibly for a variety of validation scopes, from simple ADAS functions up to complex AD systems. Furthermore, standardized formats for sensor and vehicle bus data support consistent data management and efficient test workflows.

Efficiency gains in development and validation

Development and validation teams reap the benefits: Real-world scenarios can be played back in the test lab in a deterministic manner, reproduced as often as needed and analyzed in detail. This helps to narrow down faults, shorten test cycles and make validation processes more efficient.

SensFlow can easily be integrated into existing toolchains allowing for seamless incorporation into available development and test environments. Combined with the simulation software CarMaker, the recorded data can additionally be connected with virtual environments and expanded for closed-loop testing. This hence creates a scalable basis for reproducible development and validation of modern ADAS and AD systems.