An intelligent 3D camera integrating edge computing functionality

Abstract

The utility model belongs to 3D camera technical field discloses an integrated edge calculation function's intelligent 3D camera, including 3D camera module, embedded system and communication module, embedded system integrates CPU, GPU and NPU's processor, wherein CPU is responsible for scheduling, data acquisition and transmission, GPU is responsible for point cloud generation, NPU is responsible for edge calculation, through the collection of three -dimensional data inside the camera, point cloud generation and edge calculation, only the result transmission to external device after processing, significantly reduce data transmission and the processing burden of external device. The utility model compact structure, powerful, has provided efficient edge calculation solution for industrial field, can promote system efficiency and response speed, be applicable to the defect detection, size measurement, robot navigation etc. Variety of scene, have remarkable practical value and popularization prospect.

Description

Technical Field

[0001] This utility model belongs to the field of 3D camera technology, specifically relating to an intelligent 3D camera with integrated edge computing function for industrial applications, and more specifically, to a portable intelligent device that can complete three-dimensional data acquisition, point cloud generation and edge computing inside the camera. Background Technology

[0002] With the rapid development of Industry 4.0 and intelligent manufacturing, 3D cameras are increasingly widely used in industrial settings, such as object detection, dimensional measurement, defect identification, and robot navigation. Existing 3D cameras primarily operate in two ways: one involves transmitting the acquired raw 3D data to a PC via a communication interface, where the PC generates point clouds and performs subsequent processing; the other integrates a simple embedded computing board within the 3D camera to generate point clouds, which are then transmitted to a PC for post-processing. Both methods exhibit certain limitations in industrial applications.

[0003] For the first approach, the large volume of raw data acquired by 3D cameras (such as depth maps or disparity maps) requires significant network bandwidth for transmission to the PC. This is particularly problematic in scenarios with multiple cameras operating in parallel or high real-time requirements, easily leading to data transmission delays or even network congestion. Simultaneously, the PC needs to perform complex point cloud generation and post-processing on the raw data, significantly increasing the computational burden. If further edge computing functions such as AI recognition or feature extraction are required, the PC needs additional high-performance hardware, such as a dedicated graphics card or a dedicated AI accelerator. This not only increases system costs but also adds complexity to design and deployment.

[0004] The second approach, while integrating an embedded computing board within the camera to generate point clouds and alleviate some transmission burden, has limited computing power. It can only generate point clouds and cannot perform further edge computing tasks such as AI recognition or real-time analysis. Actual needs in industrial settings often require real-time processing of point cloud data, such as target classification, defect detection, or path planning. However, existing 3D camera embedded computing boards lack efficient AI computing units, making it difficult to meet these demands. Ultimately, subsequent processing still relies on a PC, significantly diminishing the reduction in data transmission volume and computational burden.

[0005] Furthermore, the existing division of labor between 3D cameras and PCs in current technologies results in low overall system efficiency. In data-intensive applications, frequent data transfers and centralized processing not only increase power consumption but may also affect response speed due to PC performance bottlenecks, making it difficult to meet the low latency and high reliability requirements of industrial environments. Existing solutions have significant shortcomings in terms of functional integration, data processing efficiency, and system portability, urgently requiring an innovative technological solution to improve them. Utility Model Content

[0006] The purpose of this invention is to overcome the shortcomings in the above-mentioned background technology and provide an intelligent 3D camera with integrated edge computing function. By effectively combining a high-performance embedded system with a 3D camera, it realizes the integrated processing of three-dimensional data acquisition, point cloud generation and edge computing, solves the problems of large data transmission volume, heavy processing burden of external equipment and single function of existing 3D cameras, and improves the application efficiency and overall system performance in industrial field.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A smart 3D camera with integrated edge computing capabilities includes:

[0009] A 3D camera module is used to acquire three-dimensional data of objects;

[0010] An embedded system connected to the 3D camera module, the embedded system including a processor integrating a CPU, GPU and NPU;

[0011] The communication module is used to transmit the processed data to external devices.

[0012] Preferably, the CPU is responsible for scheduling, data acquisition and transmission, coordinating the workflow of the entire system, and ensuring the smoothness and stability of the data flow.

[0013] Preferably, the GPU is responsible for generating point clouds from the acquired 3D data, using its parallel computing capabilities to efficiently process depth data and generate high-quality point clouds.

[0014] Preferably, the NPU is responsible for edge computing on point cloud data, including but not limited to AI recognition, feature extraction or data compression, to meet the diverse needs of industrial sites.

[0015] Preferably, the 3D camera module can operate based on structured light, binocular vision, or TOF (time-of-flight) principles to adapt to different application scenarios.

[0016] Preferably, the communication module supports wired or wireless communication methods, such as USB, Ethernet or Wi-Fi, to facilitate flexible connection with external devices.

[0017] Preferably, the embedded system is connected to the 3D camera module via a high-speed data interface, such as a MIPI or LVDS interface, to ensure efficient and real-time data transmission.

[0018] The working principle of this utility model is as follows: the 3D camera module acquires the three-dimensional data of the object and transmits it to the embedded system; the CPU in the embedded system schedules and manages the data stream, the GPU generates point clouds from the three-dimensional data, the NPU performs edge computing on the point cloud data, and generates the processed results; the communication module transmits the processing results to external devices for user use or further analysis.

[0019] Compared with the prior art, the present invention has the following significant advantages:

[0020] 1) Reduced data transmission: By completing point cloud generation and edge computing within the camera, only the processed, simplified data is transmitted, significantly reducing bandwidth usage. 2) Reduced burden on external devices: Edge computing moves most processing tasks forward to the camera, reducing the computational load on the PC and simplifying system design. 3) High functional integration: Integrating data acquisition, point cloud generation, and edge computing functions meets diverse industrial needs and enhances equipment usability. 4) Fast response: Localized processing reduces data round-trip time, improves system real-time performance, and adapts to highly dynamic scenarios. 5) Cost-effective optimization: No need for additional high-performance hardware on the PC, reducing overall system cost and offering good economic benefits. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the structure of an intelligent 3D camera integrating edge computing function according to this utility model;

[0022] Figure 2 This is a schematic diagram of the working process of this utility model.

[0023] In the diagram: 10 - 3D camera module; 20 - Embedded system; 30 - Communication module. Detailed Implementation

[0024] The technical solutions of the present utility model will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present utility model, and not all embodiments. Based on the embodiments of the present utility model, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the present utility model.

[0025] Please see Figure 1 The present invention provides an intelligent 3D camera with integrated edge computing function, including a 3D camera module 10, an embedded system 20 and a communication module 30.

[0026] The 3D camera module 10 is the core sensing component of the entire system, used to acquire three-dimensional data of objects. In this embodiment, the 3D camera module 10 adopts the Time-of-Flight (TOF) principle, calculating the object's depth by emitting infrared light pulses and measuring the time of flight of the light pulses, generating a high-precision depth map. The 3D camera module 10 includes an infrared emitter, a photosensor, and an optical lens, featuring a compact structure, low power consumption, and suitability for continuous operation in industrial environments. It should be noted that, depending on specific application requirements, the 3D camera module 10 can also be implemented using structured light or binocular vision principles; the specific type is not a limiting condition of this utility model.

[0027] The embedded system 20 is connected to the 3D camera module 10 via a high-speed data interface. In this embodiment, a MIPI interface is used to ensure low latency and high throughput of data transmission. The embedded system 20 includes a processor integrating a CPU, GPU, and NPU, wherein:

[0028] The CPU is responsible for system scheduling, data acquisition and transmission. It adopts a multi-core architecture, supports multi-task parallel processing, and ensures the real-time performance of the data stream.

[0029] The GPU is responsible for point cloud generation. It uses its parallel computing capabilities to triangulate the depth map and generate a 3D point cloud with a precision of up to millimeters.

[0030] The NPU is responsible for edge computing, supports deep learning algorithms, and can perform AI recognition (such as target classification or defect detection) or feature extraction (such as edge contours or key points) on point cloud data.

[0031] In this embodiment, the embedded system 20 adopts a compact design and is integrated into the housing of the 3D camera module 10. The overall size is controlled within 100mm×50mm×30mm, which makes it easy to install on industrial production lines or robotic arms.

[0032] The communication module 30 is located at the output of the embedded system 20 and is used to transmit the data processed by the NPU to external devices (such as PCs or servers). In this embodiment, the communication module 30 supports a USB 3.0 interface with a transmission rate of up to 5Gbps, ensuring rapid output of processing results. Users can choose Ethernet or Wi-Fi as alternative communication methods according to their actual needs, and the specific form of the communication module 30 can be flexibly adjusted.

[0033] Please see Figure 2 The working process of this utility model is as follows:

[0034] Step 1: The 3D camera module 10 starts up, acquires the three-dimensional data of the target object, generates a depth map, and transmits it to the embedded system 20 through the MIPI interface;

[0035] Step 2: The CPU within the embedded system 20 receives the depth map, performs data preprocessing, and allocates tasks.

[0036] Step 3: The GPU generates point cloud data from the depth map, producing 3D point cloud data;

[0037] Step 4: The NPU performs edge computing on the point cloud data, completes AI recognition or feature extraction according to the preset algorithm, and generates the processing results;

[0038] Step 5: The communication module 30 transmits the processing results to an external device via the USB interface for user viewing or subsequent use.

[0039] In this embodiment, the intelligent 3D camera is applied to an industrial defect detection scenario. The 3D camera module 10 acquires three-dimensional data of the workpiece surface. After the GPU generates a point cloud, the NPU executes an AI algorithm to identify defects such as surface scratches or dents, and marks the defect locations. Finally, the defect coordinates and classification results are output through the communication module 30. The entire processing is completed inside the camera, with a response time of less than 50ms and a data transmission volume reduced by more than 80% compared to traditional methods, significantly improving detection efficiency.

[0040] This utility model features a compact structure and powerful functions, providing an efficient edge computing solution for industrial sites. It can improve system efficiency and response speed, and is applicable to various scenarios such as defect detection, dimensional measurement, and robot navigation. It has significant practical value and promising prospects for promotion.

[0041] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the present invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart 3D camera with integrated edge computing functionality, characterized in that, include: A 3D camera module (10) is used to acquire three-dimensional data of an object; An embedded system (20) is connected to the 3D camera module (10), the embedded system (20) including a processor integrating a CPU, GPU and NPU; The communication module (30) is used to transmit the processed data to an external device.

2. The intelligent 3D camera with integrated edge computing function according to claim 1, characterized in that, The CPU is responsible for scheduling, data acquisition, and transmission.

3. The intelligent 3D camera with integrated edge computing function according to claim 1, characterized in that, The GPU is responsible for generating point clouds from the acquired 3D data.

4. The intelligent 3D camera with integrated edge computing function according to claim 1, characterized in that, The NPU is responsible for edge computing on point cloud data.

5. A smart 3D camera with integrated edge computing function according to claim 4, characterized in that, The edge computing includes AI recognition or feature extraction.

6. A smart 3D camera with integrated edge computing function according to claim 1, characterized in that, The 3D camera module (10) operates based on the principles of structured light, binocular vision, or TOF.

7. A smart 3D camera with integrated edge computing function according to claim 1, characterized in that, The communication module (30) supports wired or wireless communication methods, including USB, Ethernet or Wi-Fi.

8. A smart 3D camera with integrated edge computing function according to claim 1, characterized in that, The embedded system (20) is connected to the 3D camera module (10) via a high-speed data interface, which includes a MIPI or LVDS interface.

Images