Different stages of software development for automated driving require different approaches. In all phases, simulation helps to efficiently stimulate and scale development and test activities. Functional tests in the early development phase require strictly defined scenarios with deterministic actor behavior. In later development phases, large-scale and free-running endurance tests with unpredictable edge cases become more important. AAI Replicar combines all of AAI's key technologies - from efficient, deterministic scenario tests to randomly generated scenarios in virtual endurance runs, as well as a modern user interface that allows remote access for test management.
With our naturalistic traffic simulation, it is possible to test automated driving algorithms based on long haul test drives in simulation. The test results will uncover unforeseen edge cases as well as indicate the robustness of the algorithms under test.
Automated analysis for critical situations provides rapid feedback to users and the ability to extract these situations as scenarios for later use.
AAI also produces maps in OpenDRIVE format for the simulation of test drives based on real road geometries. Please contact us for further information.
Scenario-based testing in simulation is the most suitable method for executing large numbers of known test cases. The individual scenario files allow a deterministic description of each test case as well as test automation to identify critical thresholds before validation with field operational tests.
We support the OpenSCENARIO standard and help to expand individual scenarios into parameterisable templates on request.
As an additional source for scenario files based on real situations, AAI offers the services for cloning and extracting scenarios from real test drive data. Learn more here!
Our sensor simulation is based on the pyhsical properties of each sensor type, with basic generic sensor models available for adaption towards specific sensor models. All camera imagery created can be directly delivered as segmented images.
For further information on the sensor simulation, please check our detailed introduction here.