• Realistic test of perception functions with recorded real-world data
• Extensive support of raw sensor data interfaces and buses
• Modular and easily extendable replay system to address current and future requirements 

Task

Ensuring the safety and accuracy of camera, lidar, and radar sensors under all conditions is crucial for ADAS/AD. Each sensor in the vehicle must be validated with real-world data to test its perception capabilities in different traffic and environment scenarios. Whether for sensor development, validation, or benchmarking, a robust data replay solution is required.  
 

Challenge 

The sensor market for automotive applications has become very dynamic, so suppliers must frequently come up with innovative sensor versions that support advanced communication standards and high resolutions. Perception functions which are mostly based on computer vision algorithms and which turn the sensor into a smart component must be thoroughly tested using data from real test drives. This requires data replay test systems that are capable to cope with current und feature requirements while providing enough space for flexibility and extensions.
 

Solution

The dSPACE data replay solution for validation of smart sensor components is based on a modular approach. A combination of dedicated sensor data injection interfaces and real-time capabilities ensures highly realistic data replay.

The recorded and annotated sensor data is usually downloaded to a local cache in advance or directly streamed to the replay PC from the cloud. RTMaps, the high-performant execution framework for multisensor applications, runs on the replay PC to read the recorded data streams within a wide range of file formats, such as MDF4, rosbags, or PCAP, and to push the data into the data injection interfaces, the Sensor Data Interface (SDI) Unit, or the SCALEXIO LabBox. RTMaps can also be utilized to calculate KPIs (Key Performance Indicators) after the replay by comparing the output of the sensor component with reference data. The whole data replay system can be remotely operated and monitored using gRPC and XIL API.
 

The compact FPGA-based SDI Unit for injection of raw sensor data into the smart sensor component supports relevant interfaces like FPD-Link and GMSL and is adaptable with new sensor interfaces. The SCALEXIO LabBox real-time system maintains low-jitter real-time replay of frame-based sensor data such as lidar point cloud and establishes the required restbus simulation to provide vehicle status info and keep the sensor component at the expected operating point during the whole reply. The SCALEXIO LabBox supports Ethernet (UDP and TCP/IP) and CAN/CAN FD bus networks. In a case where more advanced use cases like data fusion will need to be addressed in the future, the setup can be easily extended by adding more bus channels and compact SDI units or a modular Environment Sensor Interface (ESI) unit to establish parallel injection of multiple heterogeneous data streams into different sensors while keeping the data replay accurate thanks to hardware-based gPTP synchronization. Another advantage is that the complete system can be adapted with minimal effort for closed-loop simulation and scenario-based testing.