Ai dashcam

Abstract

The present invention discloses an AI Dashcam including an edge computing device, wherein the edge computing device includes an NPU and an AI model coupled to the NPU. The AI model and the NPU provide the edge computing for intelligence driving assistance, wherein the AI model includes an artificial intelligence driving assistance model for improving driving safety. The artificial intelligence driving assistance includes ADAS, DMS, OMS or the combination thereof.

Description

[Technical Field]

[0001] This invention relates to a dashcam, and more particularly to a dashcam with an edge computing device. [Previous Technology]

[0002] A dashcam is a device installed in a vehicle that continuously records the driver's view ahead. Some devices can also record images from the rear or sides of the vehicle, and some models support automatic uploading of pictures and videos. The development of dashcams has not only improved the accuracy of traffic accident investigations but also helped improve vehicle maintenance and fuel efficiency. Modern dashcams have become very diverse, recording not only basic vehicle data but also various other data such as vehicle position, acceleration, and braking force, and can provide more comprehensive data analysis.

[0003] One of the functions of a dashcam is to record the scene of a traffic accident for later analysis. It can also be used to assist driving safety; some models can record audio, acceleration, and GPS location information while recording video. Front-facing cameras typically have a field of view of 130 degrees, 170 degrees, or wider, and can be fixed inside the windshield, mounted on the rearview mirror, or on top of the dashboard. Rear-facing cameras are generally installed on the rear window or near the license plate, and connect to a monitor or in-vehicle screen via RCA video output.

[0004] Prior art, such as TWI823734B, announced on November 21, 2023, proposed a method for backing up dashcam data, which includes cutting background audio signal group, human voice signal group and video file into multiple fragment files, and generating a master hash value that uniquely points to the background audio signal group, human voice signal group and video file; when a preset value is determined, the fragment files are transmitted to an electronic device, and the master hash value is transmitted to a blockchain system.

[0005] Patent TWI798001B was published on April 1, 2023, proposing a method for controlling a dashcam by gesture. The method allows a camera in the car to capture images around the steering wheel at any time and transmit the captured images to a central processing unit in the car. The central processing unit recognizes a first gesture on one side of the steering wheel in the image transmitted by the camera. When it determines that the first gesture corresponds to a gear lock command, the central processing unit transmits a gear lock command to the dashcam. After receiving the gear lock command, the dashcam retains the currently recorded image file so that it is not overwritten.

[0006] Patent TWI548282B discloses a dashcam with a communication module, which includes a communication module electrically connected to an antenna receiving and transmitting unit for receiving and transmitting communication signals and establishing a network connection for the communication signals; a lens to capture images of the vehicle in motion; wherein when the microcontroller determines that the memory unit cannot store the image recording, the microcontroller uploads the image recording to the storage server through the network connection; a global satellite positioning module obtains the vehicle's positioning information and uploads it to the storage server. Patent No. TW201332822A discloses a dashcam with vehicle distance warning, lane departure warning and map display functions, including a display unit, a satellite positioning unit, an image capturing unit, a map database and a processing unit. The satellite positioning unit obtains the vehicle's location information, the image capturing unit obtains at least one continuous image of the vehicle in motion, the map database stores at least one map data, and the processing unit calculates whether the distance between the vehicle and another vehicle is less than a safe distance based on the at least one continuous image of the vehicle in motion, and determines whether the vehicle's direction of travel deviates from a lane based on a lane marking in the at least one continuous image, so as to display a reminder message and map data through the display unit.

[0007] With the popularization and development of the Internet, people's lifestyles have changed, and artificial intelligence technology is becoming increasingly mature and can be widely applied in many fields. Currently, most applications are in customer service or chat, with customer service applications using AI customer service to replace traditional customer service personnel or robots. However, many previous cases have not yet addressed integration in this area, and based on the need to expand the application level, this invention proposes a novel application field. [Summary of the Invention]

[0008] To achieve the above objectives, the present invention proposes an AI dashcam with an edge computing device, integrating intelligent services to provide safety. In one embodiment of the present invention, the edge computing device includes an acceleration circuit, which has a neural processing unit (NPU). An AI model is stored in memory and coupled to the neural processing unit. The AI ​​model and the neural processing unit provide intelligent driving assistance. The AI ​​model includes an artificial intelligence driving assistance model to improve driving safety.

[0009] According to one aspect of the present invention, the above model includes an artificial intelligence driving assistance model to assist the driver in improving driving safety and driving experience. The above artificial intelligence driving model includes one or a combination of advanced driver assistance systems (ADAS), driver management systems (DMS), and passenger management systems (OMS).

[0010] According to one aspect of the present invention, the image sensor of the AI ​​dashcam captures dynamic images of the driver and passengers, analyzes the behavior of the driver and passengers through an artificial intelligence driving model, and issues warnings or actively intervenes as needed. In other words, in one embodiment, the AI ​​dashcam of the present invention captures driving images through the image sensor and analyzes driving behavior through a driving management system; it also captures passenger images through the image sensor and analyzes passenger behavior through a passenger management system.

[0011] In another aspect of the present invention, by capturing images of the front, rear, and surrounding areas of the vehicle through sensors, and using artificial intelligence for data analysis, identification, and warning of possible errors, driving safety is improved. The advanced driver assistance system can perform one or any combination of the following tasks: monitoring road conditions, identifying obstacles, maintaining vehicle speed, maintaining lane, and maintaining distance.

[0012] According to one aspect of the present invention, the AI ​​dashcam includes a G-sensor and a communication module coupled to a processor, wherein the G-sensor includes one or a combination of an accelerometer and a gyroscope. The communication module includes one or a combination of Wi-Fi, Bluetooth, 4G, 5G, and 6G; the AI ​​dashcam includes a GPS module for positioning.

[0013] In another aspect of the present invention, AI dashcams based on edge computing can expand services and applications through AI. The AI ​​dashcam of the present invention can connect to one or a combination of user devices and in-vehicle systems. AI dashcams built on neural computing platforms can expand driving assistance, electronic map applications, and optimize navigation and positioning accuracy, etc.

Implementation Method

[0014] This invention will be described in detail here with reference to specific embodiments and their viewpoints. Such descriptions are for illustrative purposes only and are not intended to limit the scope of the invention. Therefore, in addition to the specific and preferred embodiments described in the specification, the invention can also be widely implemented in other different embodiments. The following describes the implementation of the invention through specific embodiments. Those skilled in the art can easily understand the effectiveness and advantages of the invention from the content disclosed in this specification. Furthermore, the invention can also be used and implemented through other specific embodiments, and the various details set forth in this specification can be applied based on different needs, and various modifications or changes can be made without departing from the spirit of the invention.

[0015] This invention proposes an AI dashcam 104 based on edge computing, which can be expanded with AI services and applications. According to one embodiment, referring to FIG1, the AI ​​dashcam 104 includes an edge computing device 1050 embedded in the AI ​​dashcam 104. In one embodiment, the edge computing device 1050 includes an AI model 1046 electrically connected to an AI acceleration circuit 1052, wherein the AI ​​acceleration circuit 1052 includes at least one neural processing unit (NPU) 1054. The AI ​​model 1046 is stored in memory 1053 and coupled to the AI ​​acceleration circuit 1052 to facilitate the provision of intelligent services. The edge computing device 1050 adopts a distributed computing model, transferring data processing and storage from data centers or the cloud to edge devices closer to the data source, such as the AI ​​dashcam 104 of this invention. Edge computing can reduce message latency, improve computing efficiency, and reduce bandwidth usage.

[0016] The AI ​​model 1046 has a driving assistance model that utilizes a trained artificial intelligence driving model to assist the driver in improving driving safety and driving experience. The artificial intelligence driving model can help the driver monitor road conditions, identify potential obstacles, and maintain vehicle speed and distance. The AI ​​model 1046 stores basic vehicle data, such as driving assistance model data, autonomous driving model data, and other data, and can connect with other devices, such as various sensors 1042, to help the driver obtain more comprehensive data analysis, improving driving safety and comfort. The edge computing device 1050 processes and analyzes data such as vehicle driving route, speed, and acceleration, and provides real-time feedback to help the driver better understand the vehicle's status. Sensors 1042 are located around the vehicle or integrated into the AI ​​driving recorder 104. High-resolution cameras can record images around the vehicle, providing a more comprehensive accident record. The AI ​​model 1046 uses artificial intelligence technology to analyze data through the AI ​​acceleration circuit 1052, identifying and warning the driver of potential errors, thereby improving driving safety.

[0017] The AI ​​model 1046 includes a trained artificial intelligence driving model to improve driving safety. The artificial intelligence driving model includes Advanced Driver Assistance Systems (ADAS), Driver Management Systems (DMS), and Occupation Management Systems (OMS). Advanced Driver Assistance Systems (ADAS) aim to improve driving safety and comfort. ADAS systems typically utilize various sensors 1042, such as image sensors, light, radar, and ultrasound, to monitor the vehicle's surroundings and provide real-time information and warnings to the driver, even automatically intervening when necessary. ADAS functions include lane departure warning systems, adaptive cruise control, blind spot detection systems, and parking assistance systems. DMS and OMS respectively manage and analyze the behavioral patterns of the driver and passengers, such as mental state and driving behavior. The AI ​​electronic rearview mirror 104 captures dynamic images of the driver and passengers inside the vehicle through its sensor 1042. The neural processing unit 1054 analyzes the behavior of the driver and passengers through the driver management system (DMS) and passenger management system (OMS) and issues warnings or intervenes proactively as appropriate.

[0018] The AI ​​acceleration circuit 1052 is designed to accelerate artificial intelligence computing tasks. The AI ​​acceleration circuit 1052 includes at least an NPU 1054. The AI ​​acceleration circuit 1052 improves the inference and training speed of the AI ​​model 1046 through parallel computing and specialized algorithms. To handle large amounts of data, the AI ​​acceleration circuit 1052 is typically equipped with high-bandwidth memory to ensure fast data transfer. To improve efficiency, the AI ​​acceleration circuit 1052 supports a dedicated instruction set to facilitate the optimization of AI computing tasks.

[0019] The NPU 1054 is a dedicated processing unit for artificial intelligence and machine learning, assisting in accelerating the inference of the AI ​​model 1046, improving efficiency and reducing energy consumption. The main functions of the NPU include: efficient data processing, capable of rapidly processing large amounts of data; low energy consumption, compared to traditional CPUs and GPUs, making it suitable for mobile devices and embedded systems; and optimization for specific AI and ML algorithms, enabling more efficient task execution. For object recognition and detection, the NPU 1054 can process large amounts of data from the vehicle sensors 1042 in real time, identifying and marking pedestrians, other vehicles, road signs, etc., helping the autonomous driving system make immediate decisions. The NPU 1054 can quickly calculate the optimal path and dynamically adjust according to real-time traffic conditions to ensure safe and efficient driving. In one embodiment, the NPU 1054 processes data from sensors 1042, etc., to form a comprehensive perception of the vehicle's surrounding environment. Furthermore, the AI ​​model 1046 and the NPU 1054 can analyze driving data to optimize driving strategies, improving driving efficiency and safety. In one embodiment, the NPU 1054 can process data in real time to ensure the correctness of the autonomous driving system.

[0020] The AI ​​dashcam 104 of this invention includes a processor 1041 and a sensor 1042 coupled to the processor 1041. The sensor 1042 includes one or any combination of an image sensor, radar, light, and ultrasonic sensors, used to capture information such as images, location, and distance. For example, the sensor 1042 includes a panoramic image sensor with a range of 130 to 160 degrees. The processor 1041 can convert the captured images into digital signals, compress them, and store them. A memory 1040 coupled to the processor 1041 is used to store video files, typically using an SD card or built-in memory. A G-sensor 1043 coupled to the processor 1041 is used to sense vehicle acceleration and collisions, automatically recording when a collision occurs.

[0021] In another embodiment, the AI ​​dashcam 104 may be configured to include a communication module 1045, such as one or both of a short-range communication module and a long-range communication module. In other words, the communication module 1045 includes one or any combination of Wi-Fi, Bluetooth, 4G, 5G, and 6G. The communication module 1045 is coupled to the processor 1041, allowing the user to view the recorded video and settings in real time via an application on the user device 102 (e.g., a mobile phone) through Wi-Fi or Bluetooth. The long-range communication module 1045 (e.g., 4G, 5G, 6G) is coupled to the processor 1041, allowing the user to communicate with a remote location. The communication module 1045, whether short-range or long-range, can download or update data, and can also receive dynamic information, such as weather information and traffic information. While in the vehicle or in motion, the user can connect to a remote location through the communication module 1045 to receive the required information at any time. Input unit 1047 (e.g., touch panel, microphone) and output unit 1048 (e.g., display, speaker), individually coupled to processor 1041, allow user input or user device output. In one embodiment, communication module 1045 enables edge computing device 1050 to communicate with server 106. AI dashcam 104 includes G-sensor 1043 and GPS module 1049 coupled to processor 1041. GPS module 1049 records vehicle position and speed. G-sensor 1043 includes a gyroscope and accelerometer. The gyroscope measures the angular velocity of an object, helping the device detect changes in its orientation and attitude. The accelerometer measures the acceleration of an object, detecting movement and tilt. These two types of sensors work together to provide more accurate and richer data, tracking device movement and orientation. A power supply (not shown) provides power to the dashcam, which in one embodiment may be powered by the vehicle.

[0022] Data and operating system data are stored in memory 1040. Memory 1053 and 1040 can be selected from the following or any combination thereof: including read-only memory, random access memory, non-volatile flash memory, etc. Communication module 1045 can handle signal reception, baseband processing, etc.; signals are sent to output unit 1048, such as a speaker or display. Through communication module 1045, voice, video signals, or both can be transmitted or received, or the above information can be shared with other contacts or friends in social applications, enabling the present invention to synchronously and wirelessly share driving information.

[0023] Figure 2 shows the system architecture of the present invention. The AI ​​dashcam 104 can be connected to the user device 102, and the user device 102 can also be connected to the server 106. The server 106 can be a cloud server built by others, or it can be a self-built or rented cloud server. The AI ​​dashcam 104 has an embedded or built-in intelligent model, which can improve the calculation results. The AI ​​dashcam 104 can be connected to the user device 102 through a communication network, and the AI ​​dashcam 104 can also communicate with the server 106, or indirectly connect to the server 106 through the user device 102. The aforementioned server 106 is a server under a cloud architecture. In one embodiment, the server 106 can be a dedicated server or a combination of a traditional computing server and a dedicated AI accelerator to provide higher performance and flexibility. In one embodiment, the user device 102 can be selected from smartphones 102a, tablets 102b, etc. According to an embodiment of the present invention, the AI ​​dashcam 104 can obtain the required information or update the built-in intelligent model through the server 106. According to one embodiment of the present invention, the communication network may include a LAN network, a WAN network, etc., but is not limited thereto. The communication network may be, for example, a Wi-Fi network, GSM, CDMA, TDMA, Bluetooth, or any other wireless communication protocol.

[0024] This invention utilizes the advantages of edge computing's distributed computing model to process and store data in devices closer to the data source. Specific advantages of this invention include: (1) Low latency: Since data processing is performed locally on the AI ​​dashcam 104, data transmission latency can be significantly reduced, which is highly effective for applications requiring immediate response. (2) Reduced bandwidth requirements: Processing large amounts of data on the AI ​​dashcam 104 eliminates the need to transmit large amounts of information to the cloud, thereby reducing bandwidth resources and costs. (3) Enhanced privacy and security: Data is processed and stored locally on the AI ​​dashcam 104, reducing the risk of data being intercepted or leaked during transmission. (4) Distributed processing capability: Distributing computing resources across various AI dashcams 104 improves system reliability and flexibility.

[0025] The methods or embodiments proposed above by the present invention can be executed in computer systems, servers, or similar computing systems. These server / computer systems typically include at least one controller, and peripheral devices may include storage systems, such as memory and file storage, user output interface devices, user input interface devices, and network interfaces. The network interface provides a connection interface to an external network and is coupled to corresponding interface devices of other computing devices. User input devices may include microphones, keyboards, mice, trackballs, touchscreens, touchpads, or pointing devices, and may be integrated into displays and other types of input devices. Computing devices can be of various types, including workstations, servers, computing clusters, or other data processing systems or computing devices.

[0026] The above description is a preferred embodiment of the present invention. Those skilled in the art should understand that it is used to illustrate the present invention and not to limit the scope of the patent rights claimed by the present invention. The scope of patent protection shall be determined by the appended claims and their equivalent fields. Any modifications or refinements made by those skilled in the art without departing from the spirit or scope of this patent are equivalent changes or designs made under the spirit disclosed in the present invention and should be included in the scope of the following claims. [Simplified Explanation of the Diagram]

[0027] [Figure 1] shows the functional block diagram of the AI ​​driving recorder of the present invention.

[0028] [Figure 2] shows a schematic diagram of the system architecture of the present invention.

Claims

1. An AI-powered dashcam, comprising: processor; An edge computing device is electrically connected to the processor; wherein the edge computing device includes an AI acceleration circuit, the AI ​​acceleration circuit includes a neural processing unit and an AI model stored in memory, the AI ​​acceleration circuit includes dedicated instruction set optimization computing, the AI ​​acceleration circuit improves the inference and training speed of the AI ​​model through parallel computing and specialized algorithms; and wherein intelligent driving assistance is provided through the neural processing unit and the AI ​​model, wherein the AI ​​model is a trained intelligent model, and the AI ​​dashcam can update the trained intelligent model through a server.

2. As in Request 1, the AI ​​dashcam, wherein the AI ​​acceleration circuit is equipped with high-bandwidth memory to ensure fast data transmission.

3. The AI ​​dashcam as described in Request 1, wherein the AI ​​model includes an artificial intelligence driving model.

4. The AI ​​driving recorder as claimed in claim 3, wherein the artificial intelligence driving model includes one or a combination of an advanced driver assistance system, a driver management system, and a passenger management system.

5. As in request item 4, the AI ​​dashcam captures driving images through sensors and analyzes driving behavior or status using the driving management system.

6. As in request item 4, the AI ​​dashcam captures occupant images through sensors, and the passenger management system analyzes occupant behavior or status.

7. The AI ​​dashcam as claimed in claim 4, wherein the advanced driver assistance system performs one or any combination of the following tasks: monitoring road conditions, identifying obstacles, maintaining vehicle speed, maintaining lane position, and maintaining following distance.

8. The AI ​​dashcam as described in claim 1, which includes one or a combination of an accelerometer and a gyroscope.

9. The AI ​​dashcam as described in Request 1, which includes one or a combination of Wi-Fi, Bluetooth, 4G, 5G, and 6G modules.

10. An AI dashcam as described in Request 1, which includes a GPS module.

Images