Collect, Store, Manage, and Analyze Data
for Autonomous Vehicles and Advanced Driver Assistance Systems

ADAS

Market Overview

Developing and improving autonomous vehicle systems and driver assistance systems (ADAS) requires massive real-world data from the various high-resolution cameras and vehicle sensors to train AV (autonomous vehicle) and ADAS (advanced driver assistant systems) algorithms. Obtaining this data requires purpose-built computing solutions capable of storing and recording data in real-time with speed and accuracy.

Premio offers AI Edge Inference computers that are engineered for autonomous vehicle fleets and ADAS data deployments. Our AI Edge inference computers leverage the latest technologies in compute, storage, and connectivity to accelerate machine learning and real-time inference in the harshest environmental settings on and off the road.

 What is an Autonomous Vehicle Data Computer?

A Need for Speed. Scalable Hardware Acceleration for Miles of Autonomy

Autonomous Vehicle and ADAS Data Capture and Storage Computers are designed with powerful multi-core processors, ultra-fast NVMe SSD (solid-state drive) storage, graphic engines, and high-speed connectivity to process and store incredible amounts of data generated by vehicle sensors and high resolutions cameras on the road. With more and more test vehicles being deployed, it’s critical that autonomous vehicles manufactures leverage reliable computing solutions to ensure stable development for test drives.

Power Multi-Core Processors

Ultra-fast NVMe SSD

GPU Acceleration

High speed Connectivity

Modular and Flexible EDGEBoost Nodes

Premio’s ruggedized computing solutions are capable of capturing, processing, and storing terabytes of data generated from autonomous vehicles equipped with ADAS and smart IoT sensors. These purpose-built computing solutions can assist with autonomous vehicle development for data acquisition, logging test miles and IoT senor validation for in-vehicle deployments.
Discover EDGEBoost Nodes

ADAS and Autonomous Vehicle Data Recording Computers

The primary function of ADAS and AV computing solutions is to record raw data generated from high-resolution cameras, radar, lidar, GPS, from automotive networks and fleet vehicles. Valuable data is collected, aggregated and then offloaded from the system for further analytics. Intelligent algorithms use this data to train and enable a neural network for smarter and safer vehicles. By using AI edge inference computers for data acquisition, autonomous vehicle manufactures can track and improve intelligence through the different stages of vehicle autonomy.

 Learn More About The 5 Stages Of Autonomous Driving

3 Stages in the Data Pipeline
How connected Cars and Test Vehicles Use Data

Data Ingest Stage

Consist of a mixed data from a variety of sensors on the vehicles, requiring raw storage Input/output per second (IOPS)

Data Processing Stage

The data from the ingest stage is processed for metadata tagging which requires random I/O performance from storage devices

Data Training Stage

The training stage uses data for training and relies on powerful GPUs combined with robust computing solutions that provide low latency and random read throughput from the storage media.

 Learn More About How ADAS Computers Capture Sensor Data

The Benefit for Inference and Training at the Edge

The future of autonomous vehicles requires a wealth of data. New and emerging solutions provide the necessary infrastructure for high-performance compute as well as the successful deployment for intelligent training models. AI Edge Inference computers accelerate autonomous vehicle developments by addressing challenges in data logging and acquisition, inference analysis with real-time processing, and the ability to run machine learning models at the edge.

  • Collect valuable data from devices that range from camera, radar, lidar, sonar, GPS, and CAN networks to further train neural networks
  • Deploy machine learning algorithms closer to IoT sensors for better insight and intelligence
  • Deliver low-latency results and real-time decision making with purpose-built hardware acceleration

Learn More About Inference and AI Modeling

Popular Sensor Technologies for Autonomous Vehicles and ADAS

Imaging Radar

Multi-Core Compute

LiDAR

Fully Integrated Microcontroller Units

Must-Have Requirements
For AV/ADAS Data Capture and Storage Computers

Powerful Multi-Core Processing

ADAS and autonomous vehicle computing solutions are equipped with powerful multi-core processors, providing systems with plenty of power to process and compute the Terabytes of data from a variety of sensors and high-resolution cameras.
 Learn More About Hardware Accelerators

High-Speed Storage

High-resolution cameras and sensors can generate ~4TB+ of data per vehicle per day. With such an enormous amount of data generation, vehicle computing solutions must be equipped with high capacity, high-speed SSD storage options that deliver high-performance and low latency. Modern day storage technology provide durable read/write options for a variety of random IOPS with incredible speed.
 Learn More About NVMe

Rich I/O

The computing solution must support multiple interfaces to log data from a variety of sensors. For this reason, ADAS and AV computers are equipped with plenty of I/O ports, such as USB Type-A ports, Serial COM ports, Gigabit Ethernet ports, PoE+ ports, GPIO, and video output ports, enabling connectivity to sensors, cameras, and vehicle buses and networks.
 Learn More About Industrial I/O ports

Rich Connectivity

ADAS and AV computing solutions are equipped with wired and wireless connectivity technologies, such as Gigabit Ethernet, 10 Gigabit Ethernet, Wi-Fi 6, and Cellular 4G, LTE, and 5G connectivity. Multiple connectivity options allow the computing solution to remain connected to the internet to offload mission-critical data to the cloud as well as for the system to receive over-the-air updates.
 Learn More About Wireless Technologies

CANBus Support

ADAS and AV recording computers are equipped with CANBus support to log vehicle data from vehicle buses and networks. Data logged from CANBus include vehicle speed, engine RPM, wheel speed, steering angle, and various other rich data that can provide real-time insight and valuable information about the vehicle.
 Learn More About CANBus Network And Technology

Ruggedized Design

ADAS and Autonomous Vehicle data recording computers are designed to endure challenging vehicle deployments where systems will be exposed to impacts, vibrations, extreme temperatures, dust, and other environmental challenges.
 Learn More About Essential Computing Hardware Requirements For Edge Computing

Why Choose Premio For Rugged Edge Computers

  • Premio offers purpose-built in vehicle solutions capable of capturing and storing data required for developing and improving ADAS and Autonomous Vehicle algorithms.
  • 30+ years of extensive design expertise for industrial computing solutions in x86 compute power, storage, rich I/O, and high-speed connectivity
  • Global turnkey manufacturing and support infrastructure in the USA to accelerate deployment of ADAS data capture and storage solutions and Autonomous Vehicle data capture and storage solutions.
  • Deep understanding of the computing power, storage requirements, and connectivity required for ADAS computing solutions and Autonomous Vehicle data capture and storage solutions.
  • Regulatory testing and compliance for rugged industrial computers deployed in vehicle in the North America Market.
Learn More

Frequently Asked Questions (FAQs)

Autonomous vehicles use high-resolution cameras, radar, lidar, ultrasonic sensors, GPS, and other sensors to see or perceive their surroundings. They use the sensory information to navigate themselves, avoid obstacles, read road markers such as signs in order to drive the vehicle safely. However, for the algorithm to safely guide the vehicle, huge amounts of data must be collected and stored to train AI models. The collection and storage of data require powerful, rugged computers capable of connecting to sensors and cameras, processing the data, and storing it so that it can be used at a later time to train the machine learning or deep learning models that guide autonomous vehicles.
The purpose of ADAS Recording solutions is to collect, store, and offload collected to a cental data center where the data can be used to train AI algorithms using powerful data center computers.
ADAS (advanced driver assistance systems) are, as the name suggests, electronic systems that assist drivers with driving and parking vehicles. Oftentimes, these systems are based on a software algorithm that must be trained in order to drive and park the vehicle safely. Training ADAS algorithms requires real real-world data to be fed into the neural network. The more data that’s fed to the neural network, the better it becomes at driving the vehicle.
The ADAS (advanced driver assistance systems) and AV (autonomous vehicle) data that is collected is used to train deep learning and machine learning models used to drive vehicles.
Intel estimates that an autonomous vehicle outfitted with cameras, radar, sonar, lidar, and GPUs can generate up to 4,000 GB or 4TB of data per day. The data generated by a single autonomous vehicle is more than the data generated by 3,000 in a single day.
Level 5 automation, which is the highest level of automation, requires the most amount of data because the vehicle must be able to completely drive itself without any human intervention. This requires a lot of data to train the algorithm responsible for accurately and safely driving the vehicle. Learn More About the Different Levels Of Autonomous Driving
The data recorded by ADAS computers and AV computers can be uploaded to the cloud via Ethernet, Wi-Fi 6, or Cellular Connectivity. However, since the data generated by cameras and sensors per day can reach up to 4TBs, the best solution that organizations have is to manually remove the drives and manually offload the data to a central computer systems. Manual offloading is the norm as it would take Wi-Fi or Cellular connectivity too long to upload Terabytes of information to the cloud.