New framework traces execution flows across autonomous vehicle software platforms

Modern self-driving vehicles increasingly run software stacks that blend two different middleware systems: ROS 2 (used in autonomous-driving software) and AUTOSAR Adaptive Platform. However, testing these hybrid systems has proven challenging because execution paths jump between the two domains, making it difficult to trace what's actually happening end-to-end.

The tracing problem in mixed architectures#

Scenario-based testing — running the vehicle software through realistic driving situations — is key to validating autonomous systems. But when execution flows cross from ROS 2 into AUTOSAR AP through DDS-SOME/IP middleware bridges, standard tracing tools and simulators fall short. Existing approaches can capture events from one platform or run simulations in isolation, but they cannot reconstruct the complete picture of how data flows across both systems simultaneously.

BA-TRACE brings cross-platform visibility#

Researchers have created BA-TRACE, a framework that pieces together execution flows spanning both ROS 2 and AUTOSAR AP environments. The approach combines four data sources: trace events from ROS 2, logging events from AUTOSAR's ara::log system, structural dependencies extracted from ARXML configuration files, and instrumentation added at the middleware bridge points themselves. Together, these feed into a unified end-to-end execution graph that reveals what happens as data crosses the boundary between the two systems.

Validation through a real scenario#

In a proof-of-concept study, the team applied BA-TRACE to an object-detection and emergency-braking scenario built with AWSIM and OpenSCENARIO. The framework successfully reconstructed the expected cross-platform execution path and uncovered latency hotspots specific to the boundary—such as overhead from transferring point-cloud sensor data between the systems. This visibility helps engineers spot performance bottlenecks and understand where delays originate.

Traceability vs. safety assurance#

The researchers emphasize that BA-TRACE's reconstructed topology demonstrates traceability—the ability to follow the path data takes through the system—rather than proving the software is correct or safe. It is a diagnostic tool for understanding hybrid architectures, not a safety validator. Nonetheless, this foundation for cross-platform transparency is a step toward better testing and evaluation of the increasingly complex embedded systems powering autonomous vehicles.