While discussions regarding the legal framework for autonomous transport continue in Russia, the sector itself continues to advance. Previously, we reported on the launch of autonomous trucks that are now operating on the M-11 highway and the initiation of autonomous taxi testing by Navio, the company formerly known as "Sbevravtotekh." Recently, Navio unveiled its latest innovation—the photorealistic simulator NavioSim. What does this development entail and what benefits does it promise? As a brief overview, the architecture of a typical autonomous vehicle is composed of: 1. A set of perception sensors (cameras, radars, and lidars); 2. Location-determining technologies (satellite maps and positioning modules); 3. Driving control systems (which allow maneuvering); 4. A complex software framework for decision-making. Among these components, the most complex is the software due to the substantial amount of manual programming involved. This has given rise to the Long Tail problem, making it immensely difficult to prepare algorithms for all conceivable driving conditions. To tackle this challenge, Navio has embedded generative artificial intelligence GenAI that utilizes VLA models (Vision-Language-Action), enabling a seamless integration of perception, forecasting, and actions. Despite several million kilometers tested, many potential roadway situations remain unencapsulated in the AI’s learning database. To mitigate this gap, the recently introduced NavioSim simulator aims to replicate road scenarios within a virtual reality setup. Some notable features of NavioSim include an impressive level of realism that mirrors that of standard dashcam footage and the capacity to create nearly limitless scenarios, including variations across locations and weather conditions, thereby accelerating development and enhancing safety protocols. NavioSim provides three key testing avenues: 1. SIL (Software-in-the-Loop)—quick testing of algorithm performance in a controlled format; 2. HIL (Hardware-in-the-Loop)—a method to identify issues in software and hardware interplays; 3. VIL (Vehicle-in-the-Loop)—a comprehensive virtual experience for vehicles to navigate complex scenarios that cannot be practically rehearsed in the real world. Information for these simulations is collected both from actual driving records and created using graphical editing software. It remains to be seen how quickly this technology will facilitate the integration of autonomous vehicles into regular traffic systems. Additionally, advancements are being explored in hardware solutions, including systems designed to clean the cameras of these autonomous vehicles.