The automotive industry continues to invest heavily in Advanced Driver Assistant Systems (ADAS) and autonomous driving (AD) capabilities to boost overall vehicle safety and efficiency. Before any new advanced features or capabilities can be brought to market to protect drivers and reduce accidents, they must be thoroughly vetted and tested to ensure they work reliably. This is where InoNet steps in.
With more than 25 years of experience serving the automotive industry, InoNet knows what it takes to build a computing system that can be used for testing and validating in ADAS/AD applications. The more complex the application (e.g., object detection and smart logging) the more the testing system will have to do to ensure the thoroughness and accuracy of the results. This takes collecting, storing, and processing massive amounts of sensor and camera data, in near real-time, to give car manufacturers the validation for as many scenarios and as many environments as possible.
When designing and building an in-vehicle testing system for ADAS/AD applications, InoNet has been challenged to implement storage that could meet all the demands. Most triple-level cell (TLC) storage solutions could not handle the high volume and write speeds required to collect the amount of data from all the different cameras and sensors on the car for any duration.
When storage capacity is limited, the length and distance of the test drive are also significantly reduced, necessitating the frequent return of the vehicle to home base to swap out the storage. This increases the costs of conducting the test and compromises the thoroughness and accuracy of the results, by limiting the length and variety of the test drives.
InoNet used Solidigm’s storage to build a high-performant, in-vehicle testing system that can log, store, and analyze test drive and simulation data to validate the functionality of ADAS/AD applications critical for safe vehicle operation.
As a solid-state storage (SSD) device, there are no moving parts, enabling it to maintain operations during the bumps, vibrations (0.6 G (10- 200Hz) vibration), and shocks (5G/ms shock) of test drives. The extremely efficient SSD from Solidigm in a U.2 form factor minimized the real estate, power, and cooling required, enabling the device to perform regardless of swings in temperature and humidity.
Solidigm’s drives may be small, but they are also mighty. InoNet’s hot-plug capable QuickTray® is able to store up to four SSDs, each of which can be quickly removed, swapped, and copied to get the vehicles out on the road and conducting tests faster. InoNet used four Solidigm SSD D5-P5316 Series PCIe 4.0 drives in a RAID 0 configuration to meet the high-performance requirements (continuous write speeds) of the testing system.
The InoNet Highspeed Data Logging Package uses powerful Dual Intel® Xeon® processor or AMD® EPYC™ processors, and five NVIDIA® graphics/tensor cards, making it ideal for computationally intensive AI/GPU and sensor fusion applications in the ADAS area.
The Solidigm QLC drives can store more data, more cost effectively, allowing vehicles to conduct longer, more varied test drives to improve the thoroughness and accuracy of the results. The ability to store 30% more data in the same space as traditional storage solutions makes a positive difference in the data per km that can be collected in a test drive.
Another aspect of the Car HPC Mayflower® -B17 LiQuid for cooling vehicle HPCs is the water-cooling solution which significantly enhances server performance by providing highly effective cooling for the CPU and GPUs. In addition, it consistently dissipates a stream of the heat generated to avoid throttling and improve the server performance.
The use of nonvolatile memory express (NVMe) storage technology in a QuickTray® makes the design of the testing system highly scalable to meet the needs of autonomous vehicle logging today and in the future as data demands increase.
The capacity, performance, and resiliency of the Solidigm drives, combined with their ability to obtain accurate data can be used by InoNet to support critical artificial intelligence (AI) training and evaluation, in addition to ADAS/AD testing.
For more information on how Solidigm can provide customized data storage solutions to support your cutting-edge technology requirements, please visit the Solidigm D5-P5430 page.