Autonomous Vehicle Data Computer PC
When it comes to developing, testing, and validating autonomous driving and advanced driver assistance systems, we understand that collecting, logging, and storing camera and sensor data is of the utmost importance. It’s important to have an autonomous vehicle data computer that can gather and store data generated from high-resolution cameras, radar, lidar, sonar, GPS, and other sensors because the data is vital for training software algorithms that enable ADAS and autonomous driving capabilities. Premio designs and manufactures autonomous vehicle and ADAS data computers to help organizations collect and store camera and sensor data.
Collecting and storing data is extremely important for AV and ADAS manufacturers because the more real-world data they have, the better they can train algorithms and improve their systems’ performance. Data fuels autonomous vehicles and it's a race for many autonomous manufactures to produce the best self-driving vehicles. This is so because cars must recognize every object and environmental variable they come across to drive a car safely. So, collecting as much data as possible has become a challenge for most self-driving car companies because organizations must find a computing solution capable of collecting and storing vast amounts of data generated by cameras and sensors.
In addition to being equipped with large amounts of data storage, computers that are used to collect and store autonomous vehicle (AV) data must be hardened to endure deployment in challenging environments that are not suitable for regular, consumer-grade desktop computers.
Moreover, AV data computers must be equipped with large amounts of high-speed solid-state data storage in order to store the large amounts of data produced by high-resolution cameras and sensors. Large amounts of high-speed data storage are required because AVs produce 4TB to 5TB of day data per vehicle per day.
So, having a solution that can write the data at the speed its being generated and store it is extremely important to ensure that the data is properly preserved to train machine learning and deep learning AV algorithms. The collection and storage of data is important for the makers of ADAS (advanced driver assistance systems) because it’s used to improve their algorithms and make driver assistance systems better and safer.
Premio offers autonomous vehicle data computers that are capable of connecting to the sensors and storing large amounts of vehicle camera and sensor data. Autonomous vehicle data recorders are equipped with a rich I/O, offering plenty of USB Type-A ports, Serial COM ports, RJ45 and M12 Gigabit Ethernet ports, RJ45 and M12 PoE+ ports, and a variety of other ports, enabling organizations to connect the various cameras and sensors to data computers. Furthermore, autonomous car data computers can be configured with Terabytes of solid-state data storage devices, capable of storing all of the data generated by high-resolution cameras and sensors.
Moreover, Premio’s data logger PCs are hardened for deployment in vehicles and cars that are constantly moving. Premio hardened its systems by equipping them with shock and vibration resistance, using wide temperature range components, and passively cooling them. Moreover, they are equipped with a wide power range and power protection features.
Shock and vibration resistance was achieved by creating a rugged outer enclosure made from high-quality aluminum and heavy-duty metals, protecting the sensitive internal components from vibration, shock, and impacts. Furthermore, the number of screws and joints has been reduced. By reducing them, we have reduced the number of parts that can fail, creating a more reliable and durable car computing solution.
Moreover, for added shock and vibration resistance, you have the option of configuring the system using only SSDs (solid-state drives). SSDs are better capable of operating in-vehicle environments where they are exposed to frequent and continuous shock and vibration as vehicles travel on public roads. This is so because solid-state drives store data on NAND chips, which can better handle exposure to shock and vibration than traditional hard drives. This is so because hard drives have mechanical moving parts which are physically disturbed when hard drives are exposed to shock and vibration, significantly reducing their performance. As such, to create a system that is more resistant to shock and vibration, you should configure it with SSDs. For example, some of Premio’s autonomous vehicle data PCs can be equipped up to 8x 8TB solid-state NVMe drives, providing a total of 64TB of high-speed solid-state data storage.
How Much Data Do Autonomous Vehicles and Self Driving Cars Generate?
Intel estimates that autonomous self-driving vehicles can generate 4TB to 5TB of data per vehicle per day. Self-driving cars generate so much data because they are equipped with high-resolution cameras, Lidar sensors, radar, ultrasonic sensors, GPS, and a variety of other sensors that are constantly feeding vehicle computers with data.
Image Source: Synopsys
Intel estimates that cameras generate 20-60MB/s, radar generate 10KB/s, Lidar sensors generate 10-70MB/s, and GPS generates 50KB/s, bringing the grand total to approximately 131MB/s. You have to keep in mind some vehicles are on the road for 10hrs per day. If you do the math, in a 10hr day, vehicles can generate 4.7TB of data. These numbers are not set in stone, and they will vary according to the type of cameras, sensors, and their resolution. That said, as the number of sensors and cameras on self-driving cars increases, so does the amount of data generated by vehicles.
That said, cameras and sensors are how a vehicle sees the environment around it. Without sensors and cameras, vehicles would not be able to drive themselves because they would not be able to sense the environment around them in order to avoid colliding with other vehicles, pedestrians, and objects. That said, sensors and cameras generate a ton of data that needs to be stored in order to train machine learning and deep learning models. Training vehicle algorithms is a continuous process in order to make self-driving vehicles safer than those driven by humans. The company with the most and best data will be able to develop the best AI (artificial intelligence) systems for self-driving.
What is Camera and Sensor Data Collected from Autonomous Vehicles Used For?
The camera and sensor data that is collected by autonomous vehicles is used to train machine learning and deep learning algorithms that are deployed in vehicles and cars, allowing them to identify and classify objects, such as other vehicles, pedestrians, street names, road markers, road signs, sidewalks, and many other objects that are typically encountered by vehicles. Usually, the more data used to train ML (machine learning) and DL (deep learning) models, the better the algorithm performs when presented with new environments that it has never seen before.
Using Data to Train Neural Networks - Source: Intel
Deep learning (DL) training and Machine learning (ML) training is typically performed in the cloud because it requires a lot of processing power, making data centers the best place to train them. Data centers have plenty of processing power and GPUs that can dramatically accelerate the training process. After the model is trained, it’s deployed in vehicles and cars, where it will be exposed to environments and conditions it has never seen before. Models that have been fed with enormous amounts of high-quality data will typically perform better than models that have been fed with less data.
What Are the Main Features of Autonomous Vehicle Data Computers?
1. High-Speed Solid State Storage
At this point, it should come as no surprise that autonomous vehicles generate a huge amount of data that must be collected and stored in order to improve software algorithms so that autonomous cars can better drive themselves. As such, autonomous vehicle data PCs can be configured with up to TBs of solid-state data storage, allowing organizations to store large amounts of camera and sensor data. Intel estimates that self-driving vehicles generate 4TB of data per day, having a large storage capacity is beneficial to store days worth of sensor and camera data.
That said, the amount of data storage you will need depends on how many high-resolution cameras, lidar sensors, ultrasonic sensors, and radar sensors you configure a vehicle with.
Moreover, equipping self-driving car data computers with solid-state data storage instead of hard drive data storage creates a more robust, durable, and reliable computing solution. This is so because SSDs are better capable of handling shock and vibration than hard drives since they store data on NAND chips vs the spinning metal platters that hard drives use to store data.
SSD performance is not impacted to the extent that HDD performance is impacted when vehicles are moving, exposing storage devices to shock and vibration. So, if you’re looking for a computing solution for deployment in a vehicle to store sensor and camera data, you should choose a solution that’s equipped with solid-state storage. Premio’s autonomous vehicle data capture and storage computers use SSDs since they can withstand high-levels of vibration and shock.
2. Dust and Debris Resistance
Autonomous vehicle data computers utilize a fanless design, which makes the system more durable and reliable for various reasons. First, passively cooling vehicle data storage computers makes them more reliable because it eliminates the use of fans, which are a common point of failure of many electronics, including car data computers. Second, the elimination of fans from the system eliminates the need for vents and opening. Removing all vents and openings from systems eliminates the entry of dust and other small particles into the system. This makes the system more reliable and less prone to breaking down since dust cannot damage the system’s internal components.
3. Shock and Vibration Resistance
When selecting an autonomous vehicle data computing solution, you should look for a solution that is resistant to shock and vibration. Premio’s vehicle data PCs are designed and built with shock and vibration resistance in mind. Shock and vibration resistance is achieved through a completely cableless design that eliminates the use of cables throughout the entire system. Additionally, SSDs (solid-state drives) have been used instead of mechanical hard disk drives (HDDs), making systems more reliable and robust. Lastly, Premio has reduced the number of screws and joints, reducing the number of parts that can fail, making the system more reliable.
4. Wide Operating Temperature Range
Vehicles often travel to locations that experience extreme temperatures, and so vehicle data capture computers must be able to survive extreme hot and cold temperatures while operating optimally and reliably to capture vehicle camera and sensor data. As such, vehicle computing solutions are equipped with a wide operating temperature range, ranging from -25⁰C to 60⁰C, allowing them to operate in most environments that vehicles pass through. So, whether you’re operating an autonomous vehicle that’s driving through the Mojave Desert where the temperature reaches 50⁰C or driving through New York during the freezing cold winter where the temperature reaches 60⁰C, vehicle data recording computers will operate reliably and optimally without having to invest in additional hardware to cool down the system or keep it from freezing.
5. Power Input Compatibility
Autonomous car data management systems come with a wide power input, making our systems compatible with a variety of different power input scenarios, including a car or vehicle’s battery in 24VDC or 48 VDC. Also, vehicle PCs come equipped with a variety of power protection features that include overvoltage protection, surge protection, and reverse polarity protection.
6. Power Ignition Management
Self-driving car data computers have power ignition management capabilities, enabling the computer system to detect when a vehicle has been powered on, sending a signal to the system to begin a bootup delay. Also, when the system detects that a car has been turned off, power ignition management performs a delayed shutdown of the system, allowing it to complete its tasks and shit down safely, avoiding data loss and corruption. Additionally, power ignition management prevents the vehicle computer from draining a vehicle’s power.
7. CANBus Integration
CANBus is a commonly used protocol in automobiles to deliver messages between the various components of a vehicle. Autonomous vehicle data computers can tap into the CANBus network to gather vehicle information that includes vehicle speed, engine RPM, throttle position, steering angle, tire pressure levels, and a variety of other useful information for training autonomous vehicle software algorithms.
8. Wired and Wireless Connectivity
Vehicle data acquisition and storage systems can be configured with a variety of connectivity options that include wired Ethernet, Wi-Fi, Bluetooth, Cellular, and USB. Wired ethernet is available via RJ45 or M12 Gigabit Ethernet Ports, providing blazing fast wired connectivity with routers and other devices that utilize the standard.
Wi-Fi provides ultra-fast wireless LAN connectivity where organizations can configure the connectivity speed and range according to their specific requirements. Two Wi-Fi technologies worth noting are Mu-MIMO and OFDMA. Mu-MIMO, which is short for multi-user multiple-input technologies, allows vehicle computers to connect to multiple devices simultaneously and increases a network’s throughput, making it great for high-density networks. OFDMA, which is short for orthogonal frequency division multiple access, divides Wi-Fi channels into smaller frequency allocations, allowing the system to communicate with multiple clients simultaneously.
That said, vehicles are often moving, and they may need to connect to the internet. So, vehicle data capture computer systems are equipped with cellular connectivity, allowing them to use the internet to upload critical information and download updates and other data that they may require. Cellular connectivity is available thanks to the inclusion of dual SIM sockets, allowing organizations to add two wireless data carriers for redundancy. If one of the wireless carriers is not available or has weak coverage, the device can connect to a secondary wireless carrier to offload and onload critical data.
The final connectivity option that can be added to vehicle computers is Bluetooth. Although Bluetooth does not offer the range and speed offered by Wi-Fi and cellular connectivity, it does offer reliable and simple one-to-one connectivity.
Autonomous vehicles are data-driven; they need enormous amounts of data to train and improve AI algorithms that drive the vehicle. That said, capturing and storing the large amounts of data generated by vehicle sensors and cameras presents multiple challenges for organizations. Premio offers a wide variety of solutions that can be used to gather and store autonomous vehicle cameras and sensor information. Premio’s solutions are hardened to endure deployment in challenging vehicle environments, and they are equipped with massive amounts of high-speed data storage to store vehicle data. If you have any questions about our autonomous vehicle data computers, please contact one of our vehicle computing professionals, and they will be more than happy to assist you with finding a solution that meets your specific requirements.