RoSi Simulation Platform

RoSi is a full end-to-end simulation platform, designed for the development, validation, and testing of Autonomous Driving software, especially for confined areas use cases. Thanks to highly optimized and scalable architecture, it is a perfect solution for simulating a fleet of autonomous vehicles. RoSi can be applied to productivity assessment and optimization, safety validation, and generation of synthetic data for ML models training and evaluation.

Key Features

  • Digital twin generation

    For proper simulation of the target environment, we created a semi-automatic pipeline for the creation of a digital copy of the scene based on 3D scans and architectural plans. To improve the realism of the scene, geodata, satellite images, and video recordings are used as well.

  • Realistic simulation

    RoSi simulation is based on Unity3D engine. With the created Digital Twin of the environment and simulation of a variety of sensors, test scenarios can be tested to accurately evaluate AV software before or in addition to the real-world testing.

  • Productivity/Safety metrics calculation

    To validate the system, it is critical to measure performance and quality. We provide the set of safety and productivity-related metrics which are calculated during experiments execution. We can also add more metrics if needed.

  • Intuitive GUI

    For easier use of our RoSi Simulation Platform, we created a graphical user interface for the simple creation of configuration files, experiments execution, and monitoring and analysis of experiments results.