Application Case: High-performance AI Edge Computing Empowers Vehicle-mounted Ground Truth System
In recent years, the intelligent driving industry has been booming. For intelligent driving vehicles that have been developed, comprehensive testing is required before they can be commercially produced to ensure the safety of users' lives and property. Direct road testing of test vehicles will face various problems such as safety, economy, site reliability, scene diversity, and cycle time. Virtual simulation testing can only serve as auxiliary verification, and the test results are not ideal. Therefore, there is an urgent need for a ground truth system for intelligent driving and ADAS testing to ensure the authenticity and effectiveness of the tests and assist test personnel in evaluating test vehicles.
Customer application requirements:
The vehicle-mounted ground truth system is a data acquisition system composed of millimeter-wave radar, lidar, high-precision inertial navigation, and other vehicle-mounted sensors plus efficient data recording equipment. Since it can process data to generate data (ground truth) that is more reliable than the measured sensor, the ground truth system is often used to evaluate the performance of the measured sensor. At the same time, based on the accuracy of data collection, after cleaning, labeling, and data mining of the ground truth data, a natural driving scene dataset can be formed, and then a natural driving scene library can be built.
To ensure multi-data high-synchronous collection, data collection recording, real-time data playback, efficient multi-dimensional data processing, and data verification, a high-performance, highly stable, and precise vehicle-mounted AI edge computing hardware product is required to provide strong support, with the following functions:
- High-performance processor supporting graphics cards and AI accelerator cards, open system platform, low latency, and high efficiency in processing data from vehicle perception units;
- Rich IO interfaces and powerful expansion capabilities, connecting multiple different sensor devices, and connecting to the tested vehicle communication bus;
- Multiple storage options and larger storage capacity, separate system disk and data disk, using automotive-grade solid-state storage, data disk supporting RAID0/1 function to protect data security;
- DC9-36V wide voltage power supply with reverse connection, overcurrent and overvoltage protection, adaptable to the complex and changeable power supply environment of mobile vehicles;
- Designed with a shock-absorbing chassis toenhance resistance to shock and impact, with better applicability in vehicle-mounted environments.
Related application solutions:
The vehicle-mounted solution deployed lidar, millimeter-wave radar, vehicle-mounted cameras, PTP time synchronization module, RTK inertial navigation, and vehicle-mounted Ethernet switches. JHCTECH's BRAV-7721-WP undertakes important tasks such as recording vehicle sensor data, real-time data playback, and data verification.
In this project, JHCTECH's BRAV-7721-WP adopts an Intel i9-13900K CPU, with 1*PCIe expansion slot carrying an RTX 4090 GPU, providing powerful performance to meet the performance requirements of data collection, analysis, and verification. It has strong expansion capabilities: 1*M.2 expansion 2TB NVME high-speed storage disk for high-speed data collection, using a 16TB SSD for large-capacity SSDs to meet various data recording storage needs, separate execution of system disks and storage disks to protect data security; it can support the access of 6 channels CAN-FD, including 4 channels from the German PEAK CAN-FD extended from the Mini PCIe, and the other two channels can be extended from the M.2; 2*PCIe X16 expansion slots for 2 PCIe GMSL camera capture cards to receive data from multiple cameras. In addition, the BRAV-7721-WP has multiple IO functional interfaces, including 2RS 232/422/485, 3*RJ45 Gig-LAN, 4*USB3.2, to meet the access of lidar, millimeter-wave radar, vehicle-mounted cameras, PTP time synchronization modules, RTK inertial navigation, and other sensor devices. The whole machine adopts a shock-absorbing chassis as the installation bracket to ensure the seismic effect in the vehicle's mobile environment; DC IN 9-55V wide-voltage power supply, with reverse connection, overcurrent, and overvoltage protection functions.
BRAV-7721-WP
- Intel® 12th /13th Gen Alder lake-S/Raptor lake-S CPU
- Intel® Q670 Chipset 2*DDR5 4800MHz SODIMM, up to 64GB
- 1*DP+1*HDMI and 1*VGA, UHD 8K+4K
- 2*Intel® I226V Gigabit LAN, 1*Intel® I219LM Gigabit LAN, support iAMT12.0
- 2*PCIe X16 (X8 signal)+2*PCIe X16 (X4 signal)
- 2*2.5”SATA3.0 Bays, 1*M.2 2280 PCIeX4 NVMe
- Support TPM2.0 and iVpro technology
- Fanless cooling for CPU , AI/GPU card with efficient fan cooling design
- Support dual 450W GPU + dual 75W/150W AI acceleration cards
- 1000W PSU DC IN 9-55V, and wide temperature fanless AC-DC power adapter with PFC
As a global embedded product solution provider for intelligent transportation, JHCTECH focuses on user application scenarios and vertical industries, and implements the strategic goal of scenario-enabled product innovation. In the future, we will still actively promote innovative product research and development and provide users with satisfactory product solutions.
Learn more about BRAV-7721-WP series