Validate your Sensor Simulation
Avelon
Sensor model validation
You want to use sensor models for virtual validation or to train perception algorithms? Then they need to be validated. With Persival Avelon, you can validate sensor models at any scale - from individual local test cases to continuous, server-side validation throughout the development process.
Human-Centered Interface
Avelon’s intuitive GUI makes validation accessible. It visually organizes your validation plan with block diagrams, making each step transparent and traceable. Whether you're building or reviewing, the structured layout simplifies understanding and clearly supports your validation logic — no cryptic code or guesswork involved.
Customize your Validation Campaign
Our prebuilt modules enable easy use via Drag and Drop. With an API for user-defined modules, even IP-protected content can be implemented individually using Python. The modules in different categories, such as Data Loader, Filter, or Metric, provide all the necessary functionalities to evaluate the performance of your sensor model. Thanks to the interactive and user-friendly interface, there is a short learning curve while maintaining maximum clarity.
Validation Report and CI Integration
Avelon automatically generates a PDF report with hyperlinks to navigate through your Validation Campaign. With color coding of test results and meta-information via figures and tables, the report is not only useful for sensor model developers but also for sensor model customers. The integration of Avelon within a Continuous Integration pipeline enables Continuous Validation and improves both the credibility and trust in your sensor model results.
Sensor Input Data
Avelon seamlessly handles a wide range of data formats, including images, MCAP files, rosbags, and OSI trace files. With automated calculation of filter areas around sensor model validation regions, it offers an unprecedented level of ease and user-centric design. Whether working with lidar, radar, or camera data, Avelon enables in-depth analysis across all levels of sensor outputs. It empowers users to validate not only processed and raw images, object lists, and detections, but also additional attributes like classifications.
