ADHD driving support system based on attention monitoring

Abstract

The invention relates to the technical field of intelligent driving assistance, in particular to an ADHD driving support system based on attention monitoring, which comprises a driving ability evaluation system, an emotion monitoring and intelligent driving assistance cooperation system, a natural environment emotion healing module, a traffic light waiting interval concentration training module and a driving behavior evaluation and feedback system. According to the system, the driving safety and comfort of the ADHD patient are improved through a'detection-intervention-healing 'closed-loop mechanism in combination with multi-stage intervention, emotion healing and behavior evaluation. According to the method, the problems of distraction and emotion fluctuation of the ADHD patient in the driving process can be effectively relieved, personalized support is provided, and the long-term driving habit is improved.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent driving assistance technology, and specifically relates to an attention monitoring-based ADHD driving support system and method. Background Art

[0002] In recent years, with the rapid development of road traffic safety and autonomous driving technology, the country has successively issued a series of policy documents to promote the application and popularization of intelligent driving technology. These policies provide top-level design support for improving the level of modernization of transportation and meeting the diverse travel needs of the people. However, in the process of this technological advancement, the driving safety needs of special groups such as those with Attention Deficit Hyperactivity Disorder (ADHD) have not received sufficient attention. ADHD is a common neurodevelopmental disorder. Patients often exhibit symptoms such as difficulty concentrating, impulsive behavior, and mood swings. These characteristics make them prone to problems such as distraction and untimely reaction while driving, which significantly increases the risk of traffic accidents. Although ADHD patients can legally obtain driving qualifications, society currently lacks sufficient attention to the driving difficulties of this group, resulting in a lack of research and application of driving assistance technologies for ADHD patients.

[0003] Research has shown that the high-risk behaviors and accident rates faced by people with ADHD while driving have long attracted widespread attention from the international academic community. For example, as early as 1995, foreign studies found that people with ADHD are more likely to engage in illegal behaviors and accident risks while driving, and emphasized the importance of special driving education and intervention measures for this group. In addition, research teams in Sweden and the United States, through follow-up surveys of a large number of ADHD patients, found that the risk of traffic accidents increased by 47% and 45% for male and female ADHD patients, respectively. Despite this, domestic research and policy support for driving safety for people with ADHD remain relatively weak, with a sparse number of relevant literature. Furthermore, the public generally misunderstands the driving abilities of people with ADHD, believing that they have difficulty coping with complex driving environments. This cognitive bias further exacerbates the psychological burden and reduced social participation of people with ADHD while driving.

[0004] Currently, intelligent driving assistance technology has made significant progress, especially in the areas of emotional design and emotion regulation. Leading domestic and foreign companies such as Huawei and NIO have introduced emotion recognition and intervention functions into their intelligent driving systems. By monitoring the driver's emotional state in real time and providing personalized assistance, they have improved the driving experience and safety. However, existing technologies are mainly aimed at ordinary drivers and do not adequately consider the needs of special groups, especially those with ADHD. ADHD patients not only need to deal with distraction while driving, but also need to overcome the additional challenges brought by emotional fluctuations. Existing driving assistance systems fail to fully consider these factors, lack in-depth interactive design for the emotions and behaviors of ADHD patients, and fail to effectively alleviate their driving safety issues through multi-dimensional technical means. Therefore, developing an intelligent driving assistance system specifically tailored to the characteristics of ADHD patients has important practical significance and application value.

[0005] This project aims to develop an innovative intelligent driving assistance system based on Huawei's existing ADS intelligent driving cruise and emotion recognition technologies, combined with the actual needs of ADHD patients. By building a multi-level safety assistance mechanism with a closed loop of "detection-intervention-healing", the system can monitor the driver's emotional state and attention changes in real time, and provide graded feedback and emergency response based on the degree of abnormality. At the same time, the system has also designed a natural environment emotional healing function and a concentration training module for waiting at traffic lights to help ADHD patients maintain emotional stability and focus while driving. Through the development and application of this system, not only can the driving safety of ADHD patients be significantly improved, but also personalized driving behavior assessment and improvement suggestions can be provided, filling the gaps in the current market and technology fields. Summary of the Invention

[0006] This paper addresses the problem that existing driving assistance systems fail to fully account for the distractibility, mood swings, and unexpected behaviors of ADHD patients while driving. By proposing an attention-monitoring driving support system and method, the paper proposes an ADHD driving support system and method. This system utilizes functional modules such as multi-level intervention, emotional therapy, and driving behavior assessment to form a closed-loop "detection-intervention-therapy" mechanism, enhancing driving safety and comfort for ADHD patients.

[0007] The present invention provides an ADHD driving support system based on attention monitoring, including a driving ability assessment system, an emotion monitoring and intelligent driving assistance collaborative system, a natural environment emotion healing module, a concentration training module during traffic light waiting time, and a driving behavior assessment and feedback system.

[0008] The driving ability assessment system generates assessment results of the driver's concentration and mood fluctuation trends through data collection, processing modeling and visual presentation;

[0009] Furthermore, the emotion monitoring and intelligent driving assistance collaborative system combines anomaly detection with a multi-level intervention mechanism to provide graded feedback in different driving scenarios to ensure driving safety;

[0010] In particular, the natural environment emotional healing module uses a generative adversarial network (GAN) to generate personalized natural scene images and combines them with AI-synthesized sound effects to create an immersive healing atmosphere. Furthermore, the traffic light waiting interval concentration training module uses a simple number sorting game and a dynamic difficulty adjustment algorithm to help drivers stay focused while waiting at traffic lights.

[0011] Finally, the driving behavior evaluation and feedback system comprehensively analyzes the driver's behavioral characteristics, generates a personalized intelligent driving score, and provides suggestions to improve long-term driving habits.

[0012] The specific implementation of the driving ability evaluation system is as follows:

[0013] S1. During the data collection phase, the driver's facial expressions, eye movements, reaction speed, and other behavioral characteristics are captured in real time using the onboard camera. Key point detection is performed using the OpenCV and Dlib libraries to extract key parameters such as eye position, mouth corner movement, and facial muscle changes.

[0014] S2. In the data processing stage, after standardization and regularization of the raw data, principal component analysis (PCA) is used for dimensionality reduction to screen out characteristic parameters related to attention and emotion;

[0015] In the modeling and evaluation phase, a deep neural network (DNN) is used to model static features and a long short-term memory (LSTM) network is used to predict time series features. This generates evaluation results of the driver's attention concentration and emotional fluctuation trends.

[0016] During the visualization phase, the evaluation results are displayed in charts on the dashboard. Drivers can complete the safety test process and become familiar with the system functions by watching the function introduction video.

[0017] Furthermore, the specific technical solutions of the emotion monitoring and intelligent driving assistance collaborative system are as follows:

[0018] In the anomaly detection phase, a convolutional neural network (CNN) combined with a recurrent neural network (RNN) is used to monitor the driver's emotional state in real time and predict attention change trends.

[0019] S2. Multi-level intervention stage, divided into three levels: primary intervention, intermediate intervention and advanced intervention:

[0020] -Primary intervention uses natural language generation (NLG) technology to generate short voice prompts;

[0021] -Intermediate intervention adjusts vibration intensity through an adaptive fuzzy control algorithm to provide tactile feedback;

[0022] -Advanced Intervention activates NCA Intelligent Driving Assist or Emergency Parking to take over vehicle control;

[0023] S3. In the algorithm implementation phase, the real-time monitoring part uses the YOLO model to detect facial key points and combines it with LSTM to predict time series features. The multi-level intervention part introduces a fuzzy logic rule library to dynamically adjust the prompt intensity and frequency.

[0024] In particular, the specific implementation of the natural environment emotional healing module is as follows:

[0025] In the scene generation phase, the StyleGAN2 model, a generative adversarial network (GAN), is used to generate personalized natural scene images, such as forests and beaches, based on the driver's current emotional state.

[0026] In the interactive design phase, a bilinear interpolation algorithm is used to implement image rotation and scaling, and multiple artistic filters (such as oil painting and retro styles) are supported.

[0027] S3. Rendering optimization phase, based on UE5's Nanite virtualized geometry technology and Lumen global illumination system, to ensure smooth imagery and high visual fidelity.

[0028] S4. During the immersive experience phase, AI-synthesized natural sound effects (such as birdsong and waves) are combined to create a multi-sensory healing atmosphere, helping drivers quickly calm down.

[0029] Furthermore, the technical solution of the concentration training module during the waiting period at traffic lights is as follows:

[0030] S1. During the game design phase, a simple number sorting game was developed. A set of numbers was randomly generated, and the driver had to complete the number sorting task within the traffic light countdown.

[0031] S2. Difficulty adjustment stage: Reinforcement learning algorithms are used to dynamically adjust the game difficulty to ensure the fun and effectiveness of training;

[0032] S3. Start-Stop Mechanism: Using image recognition algorithms to monitor traffic light countdowns in real time, the game is automatically started and paused.

[0033] In the feedback mechanism stage, a tactile feedback controller is combined to provide real-time vibration prompts to enhance driver engagement.

[0034] In particular, the specific implementation of the driving behavior evaluation and feedback system is as follows:

[0035] S1. Data collection phase: On-board cameras and sensors record the driver's facial expressions, eye movements, reaction speed, and other behavioral data.

[0036] In the feature extraction and modeling phase, OpenCV and Dlib were used to extract key points, and PCA dimensionality reduction was performed to retain the parameters most relevant to ADHD characteristics, thus constructing a personalized ADHD numerical model.

[0037] In the scoring and suggestion stage, an intelligent driving score is generated after each driving cycle, taking into account attention fluctuations and emotional changes, and providing improvement suggestions.

[0038] The beneficial effects of the present invention are embodied in the following ways:

[0039] The driving ability assessment system provides ADHD drivers with an accurate assessment of their attention and mood swings, ensuring they are familiar with the system's functions and adapt to personalized settings before using it.

[0040] Furthermore, the emotion monitoring and intelligent driving assistance collaborative system effectively addresses the attention distraction problem of ADHD patients in different driving situations through a graded intervention mechanism;

[0041] In particular, the natural environment emotional healing module helps drivers quickly calm down and reduce anxiety through a personalized emotional healing experience;

[0042] Furthermore, the concentration training module during the waiting period at traffic lights utilizes the idle time during driving to conduct concentration training, thereby preventing the driver from being distracted and improving the response speed of traffic lights.

[0043] Finally, the driving behavior assessment and feedback system provides targeted safety improvement suggestions for ADHD patients by tracking their driving habits over a long period of time, significantly improving their driving safety and comfort.

[0044] In summary, the present invention provides all-round driving support for ADHD patients through a closed-loop mechanism of “detection-intervention-healing”, while also providing reference value for other special driving groups.

[0045] **Brief Description**

[0046] Figure 1This is a schematic diagram of the overall system composition and structure of the present invention, showing functional modules such as single-board management, sensor data transmission, OTA updates, black box recording, and Da Vinci SDK, as well as the configuration of the Autonomous Driving Central Supercomputer (ADCSC) and multiple sensors (lidar × 3, camera × 13, millimeter-wave radar × 6, roof-mounted inertial navigation system × 1);

[0047] Figure 2 This is a trend chart of the number of domestic and foreign literature published on "ADHD driving" in the past three years, showing the changes in the number of Chinese literature included in CNKI and foreign literature included in Baidu Academic from 2022 to 2024;

[0048] Figure 3 This is a comparison chart of the experimental results of Driving-Simulation Outcomes at Baseline and at 1Month and 6Months after Training, which shows the driving simulation evaluation indicators of the intervention group and the control group at baseline, 1 month and 6 months after training, including the number of long gazes, lane position standard deviation, the number of long gazes in total g-force events, and the proportion of motor vehicle collision or near-collision events.

[0049] The accompanying drawings are numbered as follows:

[0050] 1. Single board management module; 2. Sensor data transmission module; 3. OTA update module; 4. Black box recording module; 5. Da Vinci SDK module; 6. Autonomous Driving Central Supercomputer (ADCSC); 7. LiDAR; 8. Camera; 9. Millimeter-wave radar; 10. Roof-mounted inertial navigation system.

[0051] **Specific implementation**

[0052] This invention provides an attention-monitoring-based driving support system and method for ADHD. By integrating functional modules such as driving ability assessment, emotion monitoring and intelligent driving assistance, emotional therapy in natural environments, concentration training during traffic light waits, and driving behavior assessment and feedback, it implements a closed-loop "detection-intervention-therapy" mechanism, providing comprehensive driving safety support for ADHD patients. The following describes specific embodiments of the invention in detail, with accompanying drawings and examples for illustrative purposes.

[0053] exist Figure 1The figure shows the overall system composition and structure of the present invention, which includes a single-board management module 1, a sensor data transmission module 2, an OTA update module 3, a black box recording module 4, a Da Vinci SDK module 5, an autonomous driving central supercomputer 6, and a variety of sensor configurations (lidar 7, camera 8, millimeter-wave radar 9, and a roof-mounted inertial navigation system 10). These modules and hardware devices together form the foundation of the system's operation, ensuring a complete process from data acquisition to processing modeling and intervention execution.

[0054] First, the driving ability assessment system generates an assessment of the driver's attention and emotional fluctuation trends through multi-level data collection, deep learning modeling, and visualization. During implementation, the onboard camera 8 captures the driver's facial expressions, eye movements, reaction speed, and other behavioral characteristics in real time. Key parameters such as eye position, mouth corner movement, and facial muscle changes are extracted using the OpenCV and Dlib libraries. The raw data is then normalized and regularized, and dimensionality reduction is performed using principal component analysis (PCA) to filter out characteristic parameters related to attention and emotion. These parameters are then input into a deep neural network (DNN) and a long short-term memory network (LSTM), which model and predict static and time series features, respectively, to generate an assessment of the driver's attention and emotional fluctuation trends. Finally, these assessment results are displayed graphically on the dashboard interface. Drivers can complete the safety test process by watching a function introduction video, familiarizing themselves with the system functions, and adapting to personalized settings.

[0055] Furthermore, the collaborative emotion monitoring and intelligent driving assistance system combines anomaly detection with a multi-level intervention mechanism to provide graded feedback in different driving scenarios to ensure driving safety. During the anomaly detection phase, a convolutional neural network (CNN) combined with a recurrent neural network (RNN) monitors the driver's emotional state in real time and uses a hidden Markov model (HMM) to predict attention trends. When the system detects an anomaly, a multi-level intervention mechanism is triggered. Primary intervention uses natural language generation technology to generate a short voice prompt; secondary intervention uses an adaptive fuzzy control algorithm to adjust vibration intensity and provide tactile feedback; and advanced intervention activates the NCA intelligent driving assistance or emergency parking function, taking over vehicle control. During the algorithm implementation phase, the real-time monitoring component uses the YOLO model for facial landmark detection, combined with LSTM to predict time series features; the multi-level intervention component incorporates a fuzzy logic rule base to dynamically adjust the prompt intensity and frequency. For example, in a real-world application scenario, if the driver fails to notice the vehicle ahead slowing down due to distraction, the system will first issue a voice prompt, followed by a seat vibration reminder. If the driver still does not respond, the system will automatically take over vehicle control to ensure driving safety.

[0056] Specifically, the Natural Environment Emotional Healing module utilizes the Generative Adversarial Network (GAN)-StyleGAN2 model to generate personalized natural scene images, combined with AI-synthesized sound effects to create an immersive and therapeutic atmosphere. During the scene generation phase, natural scene images such as forests and beaches are generated based on the driver's current emotional state. These images adjust brightness and color saturation based on real-time lighting conditions to enhance the therapeutic effect. During the interaction design phase, bilinear interpolation algorithms are used to implement image rotation and scaling, and various artistic filters such as oil painting and retro styles are supported. The rendering optimization phase leverages UE5's Nanite virtualized geometry technology and Lumen global illumination system to ensure smooth imagery and high visual fidelity. The immersive experience phase incorporates AI-synthesized natural sound effects such as birdsong and waves to create a multi-sensory healing atmosphere, helping the driver quickly calm down. For example, during a long drive, if a driver experiences anxiety due to traffic congestion, the system automatically triggers the Natural Environment Healing function on the central control screen, alleviating stress through soft images and soothing sound effects.

[0057] Furthermore, the concentration training module during the waiting time at traffic lights helps drivers stay focused through a simple number sorting game and a dynamic difficulty adjustment algorithm. During the game design phase, a game that randomly generates a set of numbers is developed, and the driver needs to complete the number arrangement task within the traffic light countdown. During the difficulty adjustment phase, a reinforcement learning algorithm is used to dynamically adjust the difficulty of the game to ensure the fun and effectiveness of the training. During the start-stop mechanism phase, an image recognition algorithm is used to monitor the traffic light countdown in real time and automatically start and pause the game. During the feedback mechanism phase, a tactile feedback controller is combined to provide real-time vibration prompts to enhance the driver's sense of participation. For example, while waiting for a traffic light, if the countdown exceeds 15 seconds, the system will automatically activate the concentration training game. The driver needs to click to sort the randomly arranged numbers on the screen in ascending order. The game will automatically end when the green light is about to come on, ensuring that normal driving operations are not affected.

[0058] Finally, the driving behavior assessment and feedback system provides targeted safety improvement suggestions for ADHD patients by tracking driving habits over a long period of time. During the data collection phase, the driver's facial expressions, eye movements, reaction speed and other behavioral data are recorded through the on-board camera 8 and sensors. In the feature extraction and modeling phase, OpenCV and Dlib are used to extract key points, and PCA dimension reduction is used to retain the parameters most relevant to ADHD characteristics to build a personalized ADHD numerical model. In the scoring and suggestion phase, an intelligent driving score is generated after each driving cycle, taking into account attention fluctuations and emotional changes, and providing improvement suggestions. For example, after an ADHD driver used the system continuously for one month, the system generated an intelligent driving score based on his driving data and found that his attention was frequently distracted while driving at high speeds. Therefore, it was recommended that he reduce the frequency of long-distance driving and enable the intelligent driving assistance function when necessary.

[0059] In summary, this invention achieves comprehensive driving support for ADHD patients through the synergistic effects of a driving ability assessment system, an emotion monitoring and intelligent driving assistance collaborative system, a natural environment emotion therapy module, a traffic light waiting interval concentration training module, and a driving behavior assessment and feedback system. Furthermore, the specific implementation of each functional module incorporates advanced algorithms and technical means to ensure efficient system operation and precise intervention, significantly improving driving safety and comfort for ADHD patients.

Claims

1. An ADHD driving support system based on attention monitoring, characterized in that It includes a driving ability assessment system, an emotion monitoring and intelligent driving assistance collaborative system, a natural environment emotion healing module, a concentration training module during traffic light waiting time, and a driving behavior assessment and feedback system. The driving ability assessment system generates evaluation results of the driver's concentration and emotion fluctuation trends through data collection, processing modeling and visualization. The emotion monitoring and intelligent driving assistance collaborative system combines anomaly detection and multi-level intervention mechanism to provide graded feedback. The natural environment emotion healing module uses a generative adversarial network to generate personalized natural scene images and combines AI synthetic sound effects to create an immersive healing atmosphere. The concentration training module during traffic light waiting time helps the driver stay focused through a number sorting game and a dynamic difficulty adjustment algorithm. The driving behavior assessment and feedback system comprehensively analyzes the driver's behavioral characteristics to generate a personalized intelligent driving score and provide suggestions.

2. The ADHD driving support system according to claim 1, wherein The driving ability assessment system uses an on-board camera 8 to capture the driver's facial expressions, eye movements, reaction speed and other behavioral characteristics in real time, and uses OpenCV and Dlib libraries to extract key parameters such as eye position, mouth corner movement and facial muscle changes.

3. The ADHD driving support system according to claim 2, wherein The driving ability assessment system further standardizes and regularizes the raw data, and then uses principal component analysis and dimensionality reduction to filter out characteristic parameters related to attention and emotion.

4. The ADHD driving support system according to claim 1, wherein The emotion monitoring and intelligent driving assistance collaborative system uses convolutional neural networks combined with recurrent neural networks to monitor the driver's emotional state in real time during the anomaly detection stage.

5. The ADHD driving support system according to claim 4, characterized in that The multi-level intervention mechanism of the emotion monitoring and intelligent driving assistance collaborative system includes primary intervention to generate short voice prompts through natural language generation technology, intermediate intervention to adjust the vibration intensity through adaptive fuzzy control algorithm to provide tactile feedback, and advanced intervention to activate NCA intelligent driving navigation assistance or emergency parking function to take over vehicle control.

6. The ADHD driving support system according to claim 1, wherein The natural environment emotional healing module uses the generative adversarial network StyleGAN2 model to generate images of natural scenes such as forests or beaches based on the driver's current emotional state.

7. The ADHD driving support system according to claim 6, wherein The natural environment emotional healing module further combines bilinear interpolation algorithm to achieve image rotation and scaling, and supports a variety of artistic filters.

8. The ADHD driving support system according to claim 1, wherein The concentration training module during the waiting time at traffic lights uses image recognition algorithms to monitor the traffic light countdown in real time and automatically start and pause the game.

9. The ADHD driving support system according to claim 8, characterized in that The concentration training module for waiting at traffic lights further uses reinforcement learning algorithms to dynamically adjust the difficulty of the game to ensure the fun and effectiveness of the training.

10. The ADHD driving support system according to claim 1, wherein The driving behavior assessment and feedback system uses eight onboard cameras and sensors to record the driver's facial expressions, eye movements, reaction speed, and other behavioral data. It then uses principal component analysis to reduce dimensionality and retain the parameters most relevant to ADHD characteristics to construct a personalized numerical model.

Images