Technology
Our High Performance, Efficient, and Scalable AI Model-Based Virtual Driver
At Plus, we recognize that large models are required in order to achieve the level of reasoning necessary to reach Level 4 autonomy. Our AV 2.0 approach focuses on a data-driven framework that leverages sophisticated neural network models. This model-centric strategy enables us to generalize using our vast datasets encapsulating a wide range of human driving experiences and knowledge. Through this approach, we aim to create more robust and adaptive autonomous driving systems.
Structure of Our Approach
1. End-to-end Gen-AI model for Decision Making
Our system employs advanced models to generate the best candidate driving decisions, ensuring each decision is optimized for the given circumstances.
2. Perception Layer Abstraction
Our system creates a structured abstraction that represents the real-world environment in a coherent and interpretable form. This abstraction gives us some insights on why the model would produce certain decisions.
3. Safety Envelope
To safeguard the decisions, we implement a combination of rule-based systems and neural network models, akin to a safety net, which encapsulates these decisions within a framework designed to prevent unsafe actions.
Benefits of Our Approach
Interpretability
The defined perception layer allows for clearer explanations of the autonomous system’s decisions, making it easier for engineers and stakeholders to understand and trust the technology.
Tractability
Enhanced interpretability makes it easier to trace and rectify errors or exceptions, particularly in identifying and addressing corner cases, thereby continually improving the system.
Enhanced Safety
Driving is inherently safety-critical. Our robust safety envelope, inspired by foundational safety principles, such as Asimov’s rules for robotics, ensures our system adheres to strict safety protocols.
Scalability
The flexibility of our approach allows for easy adaptation across different platforms, applications, and customer needs. This scalability is particularly valuable as it enables customers to integrate our technology according to their specific requirements, whether that involves using our complete system or incorporating individual components like the perception model.