AI Computer with Automotive-based MCU Enhancing Safety in Autonomous Driving Applications
Functional Safety in Autonomous Systems
Functional safety is paramount for both autonomous vehicle systems and robots. Achieving a safe architecture requires integrating multiple subsystems, each designed to fulfill specific roles with varying levels of complexity and numbers of components while managing and minimizing failure rate. Playing the critical role and offering performance for sensor fusion and AI inference, one must increase the computer's reliability, or reduce if not eliminate failure rate. This can be achieved by implementing redundancy. For instance, if a computer has a 1% failure rate, implementing a redundant computer architecture can dramatically reduce that failure rate down to 0.01%.
A redundant system architecture typically includes a primary computer, a secondary computer, and a safety monitor to determine which computer is the most reliable at any given moment. The challenge lies in how the safety monitor can determine whether the primary or secondary computer is more reliable and how quickly it can recognize and execute the handover.
Rugged AI Computers: Reliability Meets Performance
To enhance the safety and reliability of autonomous vehicle systems, Neousys has developed a rugged Jetson AGX Orin computer, NRU-230V-AWP. Featuring an onboard automotive MCU with an independent M12 port supporting CAN bus and digital output, it can connect to the safety monitor for real-time monitoring and predictive maintenance:
- CAN 2.0: Every second, the automotive MCU collects onboard temperature, voltage, and CPU/GPU load information, packaging them into a CAN bus message.
- Digital output (DO): Every 15 ms, the automotive MCU collects the heartbeat signal from Jetson and packages it into a voltage level: high indicates the system is active, while low indicates the system is inactive.
This monitoring enables the safety monitor to predict potential failures and initiate a seamless handover to a redundant unit within 15 milliseconds, minimizing downtime and enhancing functional safety.
By integrating an automotive-grade MCU and supporting predictive maintenance, Neousys' NRU series provide the performance and reliability critical to autonomous driving systems. Whether it's redundancy through an x86-based secondary computer, Jetson-based systems, or remote teleoperation, Neousys ensures operational safety and uptime in diverse deployment scenarios.
Neousys' Rugged AI Computers for Autonomous Driving Applications
Neousys' rugged AI computers are tailored for various autonomous applications, spanning industries such as mining trucks, agriculture tractors, last-mile delivery robots, airport ground support equipment, etc.
The computers feature:
- Rugged Designs: -40°C to 70°C fanless operation, IP66-rated waterproof, anti-shock and vibration
- GMSL2/ PoE+ GbE Camera Interfaces: Up to eight GMSL2 automotive cameras with pre-built drivers, or four PoE+ GbE
- Ready for In-vehicle Deployment: 8V to 48V DC input with built-in ignition power
- Rich Interfaces Expansion: Mini-PCIe and M.2 B-key socket for communication modules expansion
- System Monitoring Design: Supports onboard automotive-grade MCU, with digital output and CAN 2.0 interfaces in one port for potential functional safety and predictive maintenance
- Learn more about NRU-230V-AWP: https://www.neousys-tech.com/product/product-lines/edge-ai-gpu-computing/nru-230v-240s-awp
- More Neousys' Rugged AI Computers: https://www.neousys-tech.com/product/product-lines/edge-ai-gpu-computing