Cooperative vehicle infrastructure system (CVIS) -Application-MIIVII Technology

Customer Requirements

In the roadside perception scenario of CVIS, all-day and all-weather accurate perception is a difficulty. Solving this difficulty is no less difficult than assembling a perception system of an autonomous vehicle, the most important of which is the organic fusion of multiple sensors.

Solution

Based on the powerful computing of AGX Xavier, as well as the fusion perception algorithm developed by the partner Jufei, camera imaging, laser radar accurate ranging and all-weather operation of millimeter wave radar are organically integrated into the form of an all-in-one machine. In the context of smart transportation and new infrastructure, with full-speed development of the cooperative vehicle infrastructure system (CVIS), MIIVII and its partner Juefei have jointly developed a radar video all-in-one solution suitable for roadside perception in the cooperative vehicle infrastructure system (CVIS). With this solution, Jetson can effectively achieve all-round situational awareness of roads, which boosts the cooperative vehicle infrastructure system (CVIS) into the 2.0 era driven by perception.

Value

Depending on MIIVII Tech's exclusive core technology - microsecond clock synchronization based on a single-chip module, ultra-low-latency pre-fusion is performed on sensors such as cameras and laser radars. CUDA parallel computing can achieve GPU-based laser point cloud speedup for alignment and calibration of high-precision point cloud data. The errors can be greatly reduced through this multi-sensor joint calibration.

Depending on powerful software technology, this solution integrates algorithm models such as target detection & classification and semantic segmentation, and integrates high-precision lane maps, which can structure and semantically output traffic events, types and positioning information and transmit them to the cloud platform for the intelligent perception of the whole scenario; it can also communicate with a vehicle in real time through V2X for stable and real-time centimeter-level high-precision positioning of the targets.

The edge management platform and productivity tools can effectively solve the problem of "pole-climbing" deployment and maintenance in roadside perception scenarios of CVIS. This edge management platform supports remote monitoring of devices and system OTA, including remote viewing of device status abnormity warning, free upgrade and rollback of system versions through a visual interface, as well as batch burning and image configuration, thus reducing the maintenance costs for batch deployment.

Resource Center

NVIDIA

Applicant Tracking Systems