Verification and Validation of Perception Components
Machine perception components can consist of a variety of sensor types (camera types, LiDAR and radar sensors) as well as include a wide range of evaluation methods from classical image processing to artificial intelligence methods. These multi-layered solutions are sensitive to the environment and are required to measure and interpret the diversity and details of the visible environment. But how do we know if we can really trust these components in the use of highly automated systems? With novel methods we want to provide a reliable answer in exchange with different experts.
Analysis of system environment properties
We humans view and interact with the environment from our subjective perception based on our knowledge and experience. Increasingly, automated technical systems are required to replace human tasks and perform them safely in public environments. The variety of ways in which systems can sweat certain features, objects, and phenomena of urban, rural, and off-road environments is difficult to interpret from a human perspective. The goal is to structure and analyze this power of diversity to more clearly formulate more precise interpretations in the interaction of the system environment.
The maturity level of different system granularities is decisive whether a test object can be tested in simulation, on the test bench, test field or in real operation. Nevertheless, results of the test types have to be combined and transferred to a concrete question: Does the driving system meet the required specifications? Since today's driving systems have to comply with an unmanageable number of requirements, the automation of data-driven test procedures is a key element for transparent and efficient test statements.
|Front-end development for dashboard on safety topics
|Analysis of urban outdoor environments