DeepRoute.ai is making big moves in the world of self-driving cars. At IAA Mobility 2025 in Munich, the company showcased its latest platform, the DeepRoute IO 2.0. This new system is designed for mass production and aims to bring safe, smart, and adaptable autonomous driving to markets around the world. With over 100,000 passenger cars already equipped with their tech, DeepRoute.ai is well on its way to scaling up its operations.

Introducing the DeepRoute IO 2.0 Platform

The new DeepRoute IO 2.0 is powered by an innovative Vision-Language-Action (VLA) model. This technology combines visual perception, language understanding, and decision-making into a single system. It uses advanced reasoning to analyze the environment over time, making its decisions more transparent and trustworthy. The platform’s modular design allows automakers to install it across different vehicle types and brands, making it versatile for various markets and customer needs.

DeepRoute.ai is also expanding its robotaxi services using vehicles equipped with this platform. The goal is to accelerate the adoption of autonomous taxis on a large scale. The company’s dual strategy focuses on both improving its technology and entering key international markets such as Europe, Japan, and South Korea. The CEO, Maxwell Zhou, emphasizes that their VLA-powered platform is especially suited for complex European roads, thanks to its adaptability and intelligence.

The Power of the VLA Model

The core of the DeepRoute IO 2.0 is the VLA model, which integrates multiple AI capabilities. It combines visual inputs from cameras, language processing, and action planning. Enhanced with Chain-of-Thought reasoning, the VLA can understand complex situations over time and explain its decisions clearly. This transparency helps build trust with users and regulators.

By leveraging large language models, the VLA can access a broad knowledge base and learn continuously. This allows it to adapt quickly to different road signs, driving customs, and environments. It performs well even in challenging conditions, such as complex city intersections or areas with heavy traffic. Features like spatial awareness, optical character recognition (OCR), voice commands, and logical reasoning help the system respond proactively to risks and obstacles. It can execute human-like defensive maneuvers, such as early deceleration or rerouting, to keep passengers safe.

This advanced capability means the platform isn’t just reactive but proactive, making it reliable for real-world driving, especially in complex regions like Europe. The VLA’s design aims to deliver a safer, more trustworthy autonomous driving experience that can handle different driving cultures and road conditions worldwide.

Mass Production and Industry Impact

DeepRoute.ai’s latest platform supports various configurations, including vehicles with LiDAR sensors and those relying solely on cameras. It also works with multiple automotive-grade chips, giving automakers flexibility in how they implement the system. This modular approach reduces costs and simplifies integration, making autonomous driving features more accessible across different vehicle models and segments.

The company’s focus on scalability and safety is helping to push the industry forward. By combining intelligence, adaptability, and safety features, DeepRoute.ai aims to lead the next wave of autonomous vehicle deployment. As roads become more complex and demanding, such innovative platforms are essential for making self-driving cars a practical reality for everyday use.

With its strong international push and advanced technology, DeepRoute.ai is shaping the future of autonomous driving—making it safer, smarter, and more accessible for everyone. The road ahead looks promising as their solutions ready for widespread adoption continue to evolve and improve.

Inspired by

• https://ai-techpark.com/deeproute-ai-launches-mass-production-ready-io-2-0-at-iaa-2025/