With the goal of fully automated driving and the technical framework of multi-sensor information fusion, the image detection and recognition of autonomous vehicles, radar signal processing and target detection, multi-target data association and tracking, 3D point cloud scene reconstruction, vehicle fusion positioning and high-precision map, situation estimation and scene cognition, path planning and decision control technology are studied. The above technologies are integrated to develop an automatic driving system supporting multi-task modeling and reconstruction. The results have been applied in the logistics, patrol inspection and park connection scenarios of enterprise production.
The intelligent anti-fall system is mainly aimed at people who are easy to fall. The system consists of hardware and software, through the inertial sensor to identify the state of the elderly, while uploading the data to the database, and feedback to the client; when the fall state is detected, the airbag is triggered. It has a food detection function based on image processing, providing nutrition detection, diet recommendation. The hardware module adopts low-power design, uses ble Bluetooth to provide communication services, designs a fall detection algorithm with high accuracy and fast response, and provides a low-power working mode to enable the device to be used for a long time. Its food image segmentation uses artificial intelligence-based image recognition and image segmentation algorithm. Algorithm is recommended by big data and professional doctors.
1. The cloud digital twin model of the simulation platform for ship power propeller shaft is constructed, so that the cloud model can accurately reflect the real state of the physical model and effectively solve the problem of data fusion in the fault diagnosis process. 2. The fault diagnosis system of super-long shafting of ship power propeller shaft based on FPGA is developed. The intelligent IP core is used to modularize the hardware of fault diagnosis model, realize the hardware parallel acceleration processing and fast and efficient personalized correction of fault diagnosis model, and effectively accelerate the efficiency of fault diagnosis. 3. The multi-level collaborative fault diagnosis reasoning technology based on correlation model is proposed, which solves the problem of isolation of correlation fault and coupling fault of super-long shafting, and the accuracy of concurrent fault diagnosis reaches 97%.
Gesture monitoring simulation system based on flexible sensor can be widely used in various gesture monitoring and remote control, and has high application and commercial value. First, a magnetic carbon-based flexible strain sensor is used, which has high sensitivity and good stability and can effectively detect gesture motion. Besides, the sensor has the functions of electromagnetic shielding and surface hydrophobicity, which can be applied to complex application scenarios to effectively prevent external interference and water erosion. Secondly, a multi-degree-of-freedom manipulator is adopted, which is made by 3D printing and has high flexibility and plasticity. The user can control the trajectory and posture of the manipulator through gestures to achieve simulation and operation. Finally, the system is also equipped with a host computer software, which sends the actions monitored by the sensing gloves to the host computer software to display the monitoring status in real time. The user can realize gesture control and data analysis through the software, further improving the operation efficiency and accuracy
The electrochromic device with symmetrical structure is designed with transparent titanium dioxide nanocrystalline mesoporous film as the core material, which can produce blue or green gradient color change effect through voltage control, and the color change contrast can be adjusted in the range of 7%~90%. It realizes the non-polar operation of the device, and can achieve the same color change performance index under the condition of externally applied forward voltage or reverse voltage; With self-bleaching function, it can fade quickly within 1 minute without external voltage. The electrochromic/photochromic device designed by using the carbon-nitride-rich conjugated semiconductor polymer as the core material can produce the light modulation effect opposite to the conventional photochromic device, namely, the color is blue in the natural state, and the color is faded to a transparent state under the action of electricity or light. The device has the characteristics of electrical/optical dual-response, and can realize reversible color change under the control of sunlight and voltage; The special reverse color change effect makes the device produce high-resolution electro-display effect, which can clearly display various patterns and characters; It has excellent display stability and can maintain high-quality stable display effect for at least 72 hours.
The intelligent predictive control software is the fifth-generation predictive control software, which can be used not only in traditional process fields, but also in smart energy and energy Internet. It can be used in distributed energy stations, intelligent buildings and household energy consumption, wind energy photovoltaic bases, traditional processes (chemical industry, petrochemical industry, paper making, cement, thermal power, etc.) production/energy and other fields. Increasing equipment investment is not a necessary condition. It can improve the energy utilization efficiency and new energy utilization ratio, make the device/energy system run economically, improve the running stability of the device/system, improve the economic benefits and enhance the user experience.
Aiming at the major demand of equipment wear monitoring, the new working principle of optical, mechanical, electrical and magnetic four-in-one is adopted, and the lubricating oil viscosity and moisture sensors are integrated to develop a multi-element online intelligent monitoring system for equipment wear. Related core sensors, embedded controllers, monitoring software and data processing algorithms are original achievements, forming core intellectual property rights. The main functional indicators of the system include: (1) Wear particle recognition accuracy: 5μm; (2) Quantitative characterization of wear severity; (3) Real-time monitoring of trace moisture and viscosity of lubricating oil; (4) Wear real-time monitoring, trend prediction and abnormal alarm; (5) In-situ analysis of wear multivariate information based on deep learning and edge computing; (6) Multi-monitoring device networking communication and monitoring data interaction.
By using the vehicle-road coordination system, the algorithms such as vehicle license plate recognition, vehicle tracking, parking space status monitoring, indoor and outdoor integrated positioning and navigation, illegal parking monitoring, etc. are studied to solve the problems such as parking difficulty, car finding difficulty, wiping and chasing blame, parking space sharing difficulty, etc. The parking space is reserved through APP, and then parking navigation is realized. After arriving at the destination, the parking space ground lock is unlocked and parking is in place.
Based on "Bluetooth + Inertial Navigation +GNSS+C-V2X", the indoor positioning and navigation technology is researched and designed to solve the problem of high-precision continuous positioning indoors and outdoors, and realize intelligent parking and car finding service in indoor large parking lots.
Research and design model and algorithm for identifying abnormal driving behaviors based on traffic environment, vehicle movement state, driver behavior and other multi-source fusion, which can solve the problem of accuracy of abnormal driving behavior recognition and judgment during vehicle driving, and can be used in driving assistance safety systems of commercial vehicles and passenger cars.
Aiming at the real-time collaborative timing problem of multiple intersections in the vehicle-road collaborative environment, a reinforcement learning scheduling model for cloud-road collaboration is designed to control the real-time traffic timing of multiple intersections on the main road in a collaborative way, realize the green wave band traffic at multiple intersections and improve traffic efficiency.
As one of the key technologies in the field of positioning and navigation, the entangled optical quantum positioning technology uses the characteristics of quantum entangled state and squeezed state to locate the target. Compared with the traditional wireless positioning technology, the entangled optical quantum positioning technology can effectively solve the problem that the positioning accuracy is limited by the shot noise limit due to the limitation of signal bandwidth and power. In theory, it has the advantages of high accuracy, anti-interference and high confidentiality, and is not limited by spatial span. By transmitting and receiving entangled photon pairs, the entangled optical quantum positioning technology uses the second-order correlation characteristics of entangled photons to count the coincidence of two entangled photons collected in a certain period of time, and combines Einstein's time synchronization method with the strong correlation of quantum entangled sources to extract the arrival time difference of two entangled photons, so as to achieve the absolute time difference accuracy of picoseconds and the positioning accuracy of centimeters. Therefore, entangled optical quantum positioning technology is expected to subvert the traditional wireless positioning technology and become the new mainstream positioning technology in the future.
Your current location：
