Gasgoo Awards 2022 applicant: Full Stack Fusion Computing Solution from JueFX Technology

For the Gasgoo Awards 2022, Full Stack Fusion Computing Solution from JueFX Technology has applied for the Top 100 Players of China's New Automotive Supply Chain. 

Photo credit: JueFX

Product: Full Stack Fusion Computing Solution

Description

2022 is the window period of intelligent driving from L2 to L3/L4. More and more automobile manufacturers begin to lay out higher-level intelligent driving mass production, and the era of automobile intelligence has come quietly. With the improvement of LiDAR hardware technology and the decline of vehicle regulation level mass production and cost, the high-level intelligent driving function promotes the mass production of LiDAR in the field of passenger cars. For passenger cars, using LiDAR as a sensor and fusing other sensing data is a new topic, and there is still a large technological gap in the whole industry in terms of the integration of sensing and computing technologies. In order to solve the problem of mass production of high-level intelligent driving system (NOA) for passenger cars in the main engine factory, and to achieve safe and reliable high-level intelligent driving with the trend of superior LiDAR vehicles, JueFX Technology has launched the "Autopilot Full Stack Fusion Computing Solution". This scheme can provide customized front-loading and rear-loading positioning algorithms, multi-sensor fusion perception algorithms and dynamic traffic information services for different scenarios and customers, meet the needs of all stages of automatic driving, and upgrade vehicles from low-level assistant driving to NOA.

Photo credit: JueFX

Unique advantages

JueFX Technology's Full Stack Fusion Computing Solution can provide customized front-loading and rear-loading positioning algorithms, multi-sensor fusion perception algorithms and dynamic traffic information services for different scenarios and customers, meet the needs of all stages of automatic driving, and upgrade vehicles from low-level assistant driving to NOA.

The scheme can provide pure vision-based fusion perception and fusion positioning services for L2 + automatic driving system, which can adapt to mainstream vision SOC and pose sensors. For high-level automatic driving, it can also integrate Huawei MDC, Horizon J5, Nvidia Orin and other high-performance platforms, and deploy products and solutions such as fusion perception and fusion positioning according to different computing needs.

The core competence of this scheme is the fusion perception technology, which is also the core technology line that JueFX has adhered to since its establishment. The fusion perception algorithm fuses the original data of vision, point cloud and millimeter wave, and then carries out 3D tracking to output more accurate and complete perception results. At the same time, with the help of high-precision maps and road-end global samples, the behavior trajectory of traffic participants is accurately predicted.

Through extensive fusion perception practices, JueFX has developed a number of point cloud algorithms, which can enable larger models to achieve more accurate fusion perception of lidar + multi-channel vision on the embedded controller of mass production vehicles, including environmental awareness, situational awareness and behavior prediction.

JueFX Technology has also launched a high-precision fusion positioning terminal system "Minghong" for automatic driving vehicles. Its built-in environmental awareness algorithm, FPGA vehicle-level computing chip and other technical capabilities can match features with high-precision maps on the vehicle, and use visual and feature matching results to provide key information input for inertial navigation and satellite navigation. Especially in urban multi-path scenes, multi-layer overpasses or tunnels and other scenes where satellite navigation and inertial navigation are seriously interfered, it can maintain accurate and stable output of vehicle positioning.

At the same time, JueFX Technology builds a data closed-loop for the target perception and prediction of automatic driving through the integration of vehicle perception and roadside scene data model, and provides full stack technology services and empowerment for data self-updating.

Application

Commercial vehicles/passenger cars; expressway, urban roads, residential areas, etc.; vehicle lane change, long tunnel, under viaduct, AVP, etc.

Prospect

As a core technology that can significantly improve the decision-making speed and correctness of automatic driving system, "multi-sensor fusion perception" has become the top priority of industrial development. Through the core technological advantages in the integration of perception, JueFX Technology will continue to cooperate with industry colleagues to build a prosperous ecology, continue to precipitate technological strength, and help China's autopilot mass production to achieve "acceleration".