PlusAI headquarters

Three-layer redundancy architecture delivers robust physical AI for autonomous driving

PlusAI's AV2.0 Architecture for the Physical AI World

AV2.0 Three-Layer Architecture
SuperDrive perception system detecting and tracking vehicles in real-time

SuperDrive™'s perception system detecting and tracking vehicles in real-time with 3D spatial understanding

Layer 1: Primary Edge Computer

The Cutting-Edge REASONING-REFLEX Dual-Model

REASONING

Vision Language Model in action

Leverages a Vision Language Model to interpret complex real-world interactions and generate high-level driving decisions for out-of-ODD edge cases. Acts as strategic driving intelligence, understanding context and making informed decisions for unusual scenarios.

REFLEX

End-to-end model in action

Our state-of-the-art transformer-based end-to-end model fuses perception with motion planning to execute rapid, real-time maneuvers. Our network design allows us to provide industry-leading high-performance models that can rapidly scale up to the latest automotive grade SoCs.

Layer 2: Secondary Edge Computer

Redundant Fallback System

Operates on a secondary edge computer that continuously monitors the primary system's health. If a fault is detected, it triggers minimal-risk fallback maneuvers to bring the vehicle safely to a stop.

Layer 3: Cloud

Remote Operations

Provides mission control and human-in-the-loop support for scenarios beyond the virtual driver's Operational Design Domain (ODD), ensuring oversight and intervention when needed.

Key Advantages

Interpretability

Interpretability

Through semantic reasoning via a Vision Language Model and a defined perception layer, SuperDrive™'s architecture ensures that each driving decision is transparent, explainable, and trustworthy for engineers and stakeholders.

Traceability

Traceability

The architecture enables full traceability, making it possible to trace and rectify errors, identify corner cases, and support continual improvement across the entire autonomy stack.

Enhanced Safety

Enhanced Safety

Rule-based safety envelopes enable region-specific driving behavior while cloud-based remote operations provide a human-in-the-loop for scenarios beyond the virtual driver's ODD.

Scalability

Scalability

Zero-shot learning capabilities allow SuperDrive™ to adapt to out-of-sample edge cases, diverse regions, and new vehicle platforms without extensive retraining.