What is ANYVERSE?
ANYVERSE is a simulation system that improves the accuracy of autonomous vehicles in the task of object detection and classification. It consists of a scene enhancing technology and a high-fidelity render engine to produce rich and varied synthetic data, simulating the behavior of visible-light cameras and LiDAR sensors under physically-plausible scenarios.
Tagged data and LiDAR
In ANYVERSE, images are tagged automatically at pixel level, avoiding expensive and error-prone manual tagging. ANYVERSE’s ray-tracing technology can also generate LiDAR information, simulating their mechanical and physical properties, including noise.
ANYVERSE integrates seamlessly in your current training pipeline through plugins available for existing 3D platforms. Additionally, the ANYVERSE SDK provides you with the ability of developing your own custom integration.
Currently ANYVERSE supports Unreal® based pipelines.
Physically plausible variations
ANYVERSE STIR allows you to generate with minimum effort zillions variations of your base scene, such as weather, lighting conditions, and other physical plausible phenomena. STIR uses internally high-fidelity simulation engines to obtain accurate results.
ANYVERSE scales linearly in a cloud environment, dramatically reducing the time required to obtain a comprehensive and tagged synthetic data set.