Welcome: Realtimes Beijing Technology Co.,LTD.
Language: Chinese ∷  English

SUCCESS STORIES

Xavier NX+ customized carrier board based on the power grid UAV automatic fine inspection solution

The automatic flight system enables industrial UAVs to carry out power grid inspection work automatically, and the operation efficiency and inspection refinement are two times higher than the manual inspection of UAVs.

No professional transmission inspection experience or professional UAV pilot is required, which greatly reduces the labor intensity and labor time of inspection personnel. Offline 3D point cloud independently plans flight routes, sets inspection tasks freely, greatly improves the inspection frequency by image recognition technology and image control algorithm to accurately identify and lock targets, and the cloud automatically analyzes, detects defects and issues reports.


Customized power grid UAV automatic inspection based on Xavier NX

· The automatic flight system enables industrial drones to carry out power grid inspection work automatically, and the operation efficiency and inspection refinement are twice higher than the manual inspection of unmanned aerial vehicles. Greatly reduce the labor intensity and labor time of inspection personnel.

·3D point cloud independently planned flight routes and set inspection tasks freely, greatly increasing inspection frequency. Image recognition technology and image control algorithm accurately identify and lock the target, the edge of the automatic analysis, detection of defects and issue a report.

· Feiyun Intelligent box based on Jetson NX has advantages such as small size, light weight, high performance and low power consumption, which makes it suitable for landing deployment in UAV and other edge ends


Adaptive panoramic solution for large and medium-sized vehicles based on NX+ customized carrier board

The adaptive panoramic system is specially developed for the traction semi-trailer. In the vehicle driving, turning, turning, reversing, parking and other scenes, the multi-channel AHD video access provides 360-degree visual information without blind spots, multiple radar data fusion, accurate perception of the surrounding environment of the vehicle, NX high-performance intelligent analysis, with God's perspective to assist the driver to safely control the vehicle. Effectively reduce rolling, collision, rubbing, fall and other accidents (also applicable to other types of vehicles and large construction machinery).

Based on the AI vision algorithm, the image features of each camera are matched in real time to perfectly cope with the dynamic changes of the Angle between the traction locomotive and the semi-trailer, and the natural stitching and fusion effect is comparable to that of aerial photography.

With efficient calibration algorithm, minimize the impact on customer business.

Starlight sensors are not afraid of the dark, the display is also clear at night.

This case has been successfully developed and launched for domestic and foreign markets.

Intelligent control system based on Xavier+RTSO-1001

The intelligent control center based on NVIDIA Jetson AGX Xavier + RTSO-1001 effectively solves the problem that many cameras, radar and sensors import huge amounts of data, and the performance is obviously insufficient on the original computing platform. It helps colleges and universities achieve excellent results in obstacle detection, safety warning, obstacle circumnavigation and overtaking performance competitions of intelligent autonomous vehicles in China.

The application of the intelligent control center based on AGX Xavier+ RTSO-1001 in UAV helps universities to achieve outstanding results in the international UAV competition by accurately positioning and rapidly capturing targets with multiple sensors.

At present, we have exchanges and cooperation with 100+ universities based on Jetson product solutions. Help universities in visual image processing, pattern recognition, deep learning, machine vision and other aspects of the study and research.


Lidar sensing scheme based on AGX+RTSO-1001

Different from FPGA or various customized ASics, NVIDIA's Jetson AGX Xavier development platform can be regarded as a mobile computer with a perfect upper-layer system environment, which can quickly and seamlessly transfer desktop development results to it. What you see is what you get development ensures consistency between development and deployment. Eliminating the embedded migration, testing and other links accelerates product development on another level.

Thanks to the NVIDIA Jetson AGX Xavier platform with low power consumption and up to 32T computing capacity, a variety of liDAR and its combined application can realize real-time omnidirectional perception of obstacles in the surrounding environment, such as detection, classification, tracking and semantic segmentation. Giving autonomous lidar perception and analysis capabilities in a simple, fast and flexible way.

The sensing scheme greatly improves the speed of the sensing software in processing large amount of point cloud data. Support automatic driving customers to use multi-Lidar flexible combination application, small size plug and play, further enhance the product user experience, help promote the vehicle road collaboration, medium and low speed logistics robots, industrial automatic transportation, automatic driving medium and low speed scenarios and applications safely landing.


AI fundus screening machine based on NANO+ customized box

Baidu Medical's AI fundus screening machine can help patients quickly screen various fundus diseases, including sugar net, glaucoma, old yellow, and prevent the risk of blindness in advance. Its greatest significance is to expand the scope of fundus screening, which is able to help patients at a more basic level, more remote and less accessible places that ophthalmologists currently cannot reach.

5.0 technology can realize multi-level medical record structure, fully support the application of medical data in clinical, scientific research, management and other scenarios, and build a clinical assisted decision system following evidence-based medicine to improve medical quality and reduce medical risks. The system includes multiple functions, such as auxiliary inquiry, auxiliary diagnosis, treatment plan recommendation, doctor's order quality control, etc., in which the four physiological structures of fundus are extracted through the deep learning accurate algorithm, and the risk of fundus disease is assessed, and the analysis results are finally formed.

> The workflow of information collection, image analysis and local film reading can be successfully completed without network

> Realize multi-level medical record structure, fully support clinical, scientific research, management and other scenarios

> More than 1,000 medical institutions have been established in 18 provinces and autonomous regions.


Customized unmanned delivery robot based on NX+RTSO-6002

The low-speed autonomous driving system is based on NVIDIA Jetson XavierNX+6002 and adopts multi-sensor data fusion, visual pattern recognition and other technologies to realize low-speed autonomous driving. At present, the products in the field of intelligent low-speed roaming vehicle human low-speed automatic driving truly realize mass production and scenario-oriented application, which can be widely used in industrial/science and technology parks, tourist attractions, urban parks, resorts and other scenes. Based on the combination of indoor and outdoor low-speed automatic driving technology and robot chassis control technology, it is an unmanned distribution platform integrating transportation, technology, information and entertainment interaction.


Blister detection solution based on Xavier NX+ customized carrier board

With the foreign high-speed and high-precision medicine machine continues to enter the Chinese market, the domestic machinery manufacturing enterprises have also upgraded the products. The tablet pellet filling machine has fast speed, high filling accuracy requirements, and belongs to the medical machinery equipment with a relatively high degree of automation. The products from PVC film through hot pressing die to pill filling, and then to aluminum plastic sealing and pressing slice need to be completed at a high speed. This requires that there is no leakage filling of the tablet and no breakage of the tablet. Usually, after the tablets are filled, the tablets will be directly sealed with aluminum plastic. If the plastic holes are not filled or the defective tablets are filled, the product waste and the difficulty of detection will be improved. In addition, many plastic seals are opaque, and once they are well sealed, it is difficult to detect, so the quality of products will sometimes be damaged. Blister vision detection platform can detect one or more tablets in real time.

The equipment adopts advanced machine vision detection technology to detect many defects on the surface of the medicine plate, such as dirt, bubble leakage, bad heat sealing, pressure bubble, raw edge, heterochromatic bubble and so on. Supports a variety of waste output, including network port, serial port, direct I/O output, etc


Based on AGX Orin intelligent box unmanned mine card solution

At present, there are many challenges in mine, such as: poor transportation environment, high safety risk, traffic accidents; Personnel recruitment is difficult, occupational diseases; The manual driving speed is slow, the transport capacity is poor, the transportation efficiency is low, the transport mining card unmanned driving is an effective solution to the above problems, so it is an important part of the intelligent construction of open pit mine.

RTSS-Z201's rich interface and high computational power AI analysis, integration of multi-sensor accurate data, help mining card intelligent driverless solution, can help mining enterprise customers to significantly reduce labor costs, save energy consumption of mining vehicles, and increase the overall annual benefit by more than 30%.


Intelligent parking system based on TX2+ customized load board

By using NVIDIA Jetson TX2 Module to build the world's most advanced full-scene image supervision equipment "intelligent Sky Eye", it can realize all-weather 7x24 hour monitoring, and form complete image evidence according to the vehicle trajectory, reaching a system accuracy of more than 99%, with electronic payment. Realize the management mode of no charge, "no sense of payment" parking experience.


Subway passenger flow intelligent analysis and early warning system based on Feiyun Smartbox

The high-precision passenger flow sensing device directly obtains the stream information of the webcam and decodes the stream information. Combined with the NVIDIA Jetson TX2 Module, it conducts the pedestrian target detection based on deep learning. By using the association information between successive frames of images, it generates a variety of hypothesis trajectories and calculates the prior probability of each hypothesis. Combined with the apparent and motion information, the prediction was carried out, and the color information and depth information were introduced for matching to improve the correlation accuracy, realize the dynamic multi-target tracking, and output the target tracking trajectory. The statistical parameters such as human flow, speed, flow direction and density within the region or through the boundary were accurately obtained through the trajectory analysis. Finally, the obtained data information is uploaded to the database, and the update frequency of 1 second is adopted to ensure the security of the data. Combining various data resources, a big data analysis platform centering on passenger flow is established to provide comprehensive services for passenger flow safety management, risk prevention, efficiency improvement and commercial value mining.

NVIDIA Jetson TX2 Module deep learning platform is adopted to greatly improve the accuracy, computing time and efficiency of detected pedestrian data, and the accuracy of passenger flow calculation is greater than or equal to 96%. The detection accuracy of passenger flow density is greater than or equal to 95%; The average speed of passenger flow detection accuracy is greater than equal 92%; The accuracy rate of pedestrian tracking trajectory is greater than or equal to 95%.


close
Scan the qr codeClose
the qr code