Overview
As autonomous mobility reshapes urban transportation, edge computing has become the backbone of real-time decision-making inside self-driving vehicles. A global autonomous mobility solution provider developing Level-4 self-driving shuttles faced growing challenges around onboard AI processing, thermal constraints, and system reliability across diverse climates. Premio’s VCO-6000-RPL Series, GPU-capable machine vision computer delivered the performance, scalability, and durability required to support mission-critical autonomous operations worldwide.
Challenges
- Insufficient CPU and GPU computing power to process multi-sensor data in real time
- Limited support for full-size, high-performance GPU expansion within a compact vehicle enclosure
- Lack of native high-speed networking and CANbus connectivity for autonomous vehicle systems
- Thermal stress caused by enclosed in-vehicle deployment under extreme ambient temperatures
- Need for a rugged platform certified for safety and reliability in global transportation environments
Solution
- Premio’s VCO-6000-RPL Series GPU-ready machine vision computer
- Support for Intel® 12th, 13th, and 14th Gen processors on an LGA 1700 socket
- Dual 10GbE fiber networking and integrated CANbus interfaces for vehicle communication
- Wide 9–48VDC power input with dedicated 600W GPU power board
- Industrial-grade thermal design supporting extended operating temperatures
Benefits
- Reliable real-time AI processing onboard autonomous shuttles
- Stable performance across extreme global temperature conditions
- Simplified system integration for scalable fleet deployments
Company Overview
The company designs and deploys Level-4 autonomous electric shuttles for last-mile and on-demand public transportation. Its end-to-end approach spans vehicle engineering, autonomous system integration, and fleet operations for smart mobility ecosystems. With deployments across Europe, North America, Asia, and the Middle East, the organization continues to expand sustainable autonomous transport worldwide.
The Challenges
Real-Time Sensor Processing Demands
Autonomous shuttles rely on continuous streams of data from cameras, LiDARs, radar, and onboard sensors. Processing this information in real time requires substantial CPU and GPU resources to ensure safe navigation and decision-making. Existing platforms struggled to deliver consistent performance under these workloads.
GPU Expansion Constraints Inside Vehicles
Deploying AI workloads onboard required a full-size, high-performance GPU, but space and power limitations inside the vehicle created significant integration challenges. Many computing platforms could not support the physical size, power draw, or thermal needs of modern GPUs. This limited the ability to scale AI capabilities across the fleet.
Vehicle-Native Connectivity Requirements
Autonomous systems depend on fast, reliable communication between compute, sensors, and vehicle subsystems. The platform needed native CANbus support and high-speed networking to maintain deterministic data flow. Without these interfaces, system latency and integration complexity increased.
Extreme Thermal Conditions
Mounted inside enclosed electric shuttles, the computing system faced sustained exposure to high internal temperatures. Vehicles operating in regions such as the Middle East and southern Europe further amplified thermal stress. Conventional systems risked throttling or failure under these conditions.
Global Compliance and Reliability Expectations
Operating across multiple regions required compliance with safety and industrial standards. The computing platform needed to meet rigorous certification requirements while maintaining long-term reliability in mobile, vibration-prone environments.
The Solution
GPU-Ready Industrial Machine Vision Platform
Premio’s VCO-6000-RPL Series was selected as the core onboard computing platform for the autonomous shuttles. Its rugged construction and expansion-ready design allowed seamless integration inside vehicle enclosures. The system provided a stable foundation for real-time autonomous AI workloads.
High-Performance Intel® Processing Architecture
Equipped with an LGA 1700 socket supporting Intel® 12th, 13th, and 14th Gen processors, the system delivered the computing headroom required for complex perception and control algorithms. Combined with up to 64GB of DDR5 ECC memory, it ensured deterministic performance for safety-critical operations. This processing capability enabled consistent AI inference across all deployments.
Full-Size GPU Support with Dedicated Power
The platform’s dual PCIe Gen 4 x16 slots and integrated 600W GPU power board supported full-size, high-performance GPUs. This allowed the autonomous shuttles to run advanced vision and sensor fusion models directly onboard. The result was faster decision-making without reliance on external compute resources.
Integrated High-Speed Networking and CANbus
Dual 10GbE fiber networking and customized CANbus interfaces enabled reliable communication between sensors, vehicle controllers, and fleet systems. These native interfaces reduced system complexity and improved data throughput. The configuration ensured seamless integration with existing autonomous vehicle architectures.
Thermal and Environmental Resilience
Designed for operation from -25°C to 70°C, the system maintained stable performance even in harsh, enclosed vehicle environments. Its industrial thermal design prevented throttling during prolonged high-temperature operation. This reliability proved critical for deployments across diverse global climates.
The Benefits
Consistent Autonomous Performance
Real-time processing ensured safe and responsive shuttle operation across varying traffic and environmental conditions.
Scalable Global Deployments
A standardized computing platform simplified expansion into new regions and vehicle fleets.
Reliable Long-Term Operation
With industrial certifications and robust design, the system minimized downtime and maintenance overhead, supported by Premio’s engineering expertise in Los Angeles.
Conclusion
By deploying Premio’s VCO-6000-RPL Series GPU-ready industrial machine vision computer, the autonomous mobility provider achieved reliable, real-time AI processing inside its self-driving shuttles. The solution addressed performance, thermal, and integration challenges while enabling scalable global deployments. This collaboration continues to support the evolution of safe, sustainable autonomous transportation.