ZigBee-based driver fatigue warning system research and design

Abstract: The fatigue monitoring system for todays relatively simple way of monitoring, less reliable, if taken by a variety of monitoring methods and MAX306CWI datasheet and integration of automotive wiring within limited space constraints more difficult. For this series of questions, this paper presents a wireless sensor network based on ZigBee technology with sensor data fusion technology combines driver fatigue warning system, the system is responsible for the CC2430 zigbee network and MAX306CWI price and coordinate the establishment and MAX306CWI suppliers and management, high-performance processors to complete fatigue Information Fusion. After tests showed that the system is suitable for vehicle running and be able to better monitor the behavior of fatigue, reliability, up to 95%.

1 Introduction

With the rapid development of automobile industry, is also associated with rapid increase of traffic accidents, and these accidents, traffic accidents caused by fatigue driving of the total 16% of the highway is more than 20% [1], so the development of prevention and alarm device driver fatigue become the focus of the study of fatigue and difficulty driving. With the emergence and development of new technologies, this paper based on the ZigBee wireless sensor network technology and information technology integration decision-making system, the combination of fatigue driving. ZigBee technology through the formation of the network has the following characteristics: monitoring methods as a result of a single poor performance [2], while the expansion of the system and wiring by the body space, cost and other restrictions. Therefore, use of ZigBee wireless technology, security, reliability, low power consumption, combined with the SCM control technology with them to introduce the design of the sensor to the vehicle, not only eliminates the need for the installation of communication cables, reducing the amount of installation works, but also safe and reliable to achieve data transmission and networking, to develop a stronger vehicle sensor applicability. Another low-power ZigBee technology, the characteristics of ideal low-cost way to travel.

2 fatigue detection system architecture and principle

2.1 ZigBee network formation and communication

car-based short communication distance of each node, expand network coverage without the router This model uses the Zigbee star network structure, only the coordinator and the sensor device can constitute a network, thereby reducing the complexity of the whole system network. The middle of the coordinator is responsible for initiating and maintaining the network and the information collected to the high-performance processors have completed the integration of decision-making information, and then from the processor to determine the results to the fatigue warning device.

(1) lane departure detection, eye detection frequency, time of testing eyes closed, yawning start with four sensors that gather information on the original image, and then pass the information to deal with their own DSP chip to get fatigue information, the final results of the collected signal transmitted to the ZiBee SoC module. This use of TIs CC2430, CC2430 single chip RF integrated ZigBee (RF) front end, memory and microcontroller. In the receive and transmit mode, current consumption less than 27mA, respectively, or 25mA. CC2430 system requirements for the vehicle requirements of a very long battery life is more appropriate. This information last four sent to the coordinator by the CC2430. Implementation framework of the specific flow chart in Figure 1.

Figure 1, the fatigue monitoring system based on ZigBee diagram

(2) Coordinator is responsible for networking and management of the terminal sensors, network basic process: First, scan the energy scan and activate the channel, if you find the right channel, then create a unique 16-bit network PAN ID, the ZigBee network address of the Coordinator of the fixed-network short set to 0, then start broadcasting network to the surrounding information, and accepted for processing within the scope of its coverage to join the networks request, then add the new node information. Network diagram as shown in Figure 2.

Figure 2 Network diagram

flow chart we can see from the coordinator is not handled and stored The information sent by sensor nodes, which fatigue the information directly to the high-performance processor, so the coordinator can better manage their networks. Fatigue characteristics of the processor is responsible for the integration of multiple judgments. The model will be the coordinator and the signal sink node (gateway) unified design, coordinator / gateway is responsible for communication with the various terminal equipment and external network communications. If the driver fatigue, driving in serious condition, easy to cause a traffic accident when the driver information coordinator will be sent to the gateway, and then into the external network message format, and finally through the GSM / GPRS network communication mode and highway safety, sent to the remote monitoring equipment.

2.2 sensor information acquisition technology

driver face image through infrared camera and LED to obtain, which can be issued for the lighting of the LED 850nm and 950nm, respectively spectrum of light. When using a different IR, when the eye pupil will show different colors. When using 850nm infrared light illumination, the pupil showed a red, commonly known as red-eye effect; but with 950nm infrared light illumination, the pupil preached black. In addition to the pupil than two images, other parts of the face are the same. By comparing these two images can easily navigate to the eye, and then through a series of image processing to get the parameters of the face and achieve eye tracking. In addition, the use of infrared LED light to reduce interference to ensure image quality around the same time, the driver can reduce the visual interference of light as it is almost invisible. Comparison of the eye shown in Figure 3:

eye detection comparison chart in Figure 3 (a) red-eye; (b) black eye; ( c) difference image

Figure 4 IR CCD camera

the same time in order to get Figure 3 (a) , (b) two images, shown in Figure 4 can be used infrared camera device, when the incident light shine on the middle of the spectral slices (which can be separated into the incident laser line of reflection / transmission ratio of 1, two beams of light) when , can be divided into two parallel incident light beam, then through the 850nm and 950nm, respectively, the filter into the camera, so at the same time get the two images in different colors in addition to the pupil, the other is the same. For a limited time to complete a large number of image data processing, DSP image processing chip TMS320DM642, it is the image processing speed can reach more than 25 frames per second, which is complete in 40ms on an image processing operation, plus on the CCD camera is 25 frames per second PAL system, these devices enough to complete real-time image processing, speed up the completion of the pilot blink frequency, eyes closed slowly, yawning fatigue feature extraction and calculation.

Drive offset detection is based on the behavior of the vehicle response to indirect signs of driver fatigue. CCD camera will be facing the direction of vehicle travel, monitoring the direction of the vehicle, while monitoring the turn signal. If the car changes direction of the turn signal is not turned on, then that driver has driven into the possibility of fatigue. Vehicle behavior detection is not based on human performance activity, so lack of face detection can complement each other and man-made differences, while the driver was not due to fatigue but because of other factors (such as mobile phones, music, children) to determine the error caused by lack of concentration when can also give some have to remind.

2.3 Sensor Fusion

us to 120ms for a small period, because the system can be collected in a frame within 40ms of images, the latter 80ms for image processing, so that a minutes, a total of 500 images. Analysis of these images to the drivers fatigue situation. According to ergonomic principles, the body appeared tired, blinking frequency than normal in a time significantly faster, which is when the driver tried to stay awake in the fatigue response, into a deeper level of fatigue occurs when the eyes closed one time plus length features. Eyes eyes opened when awake during a closed down only a few dozen frames of the time frame (0.25 seconds), and fatigue, you need 20 or a second or two; mouth when yawning significantly increased the vertical radius. When we first collected the blinking frequency of the normal driver, his eyes closed and mouth first time, information, and then the fatigue and the occurrence of the situation when compared to judge the degree of fatigue. We use fuzzy logic method to get information on the acquisition integration decision-making. For example: When the Coordinator receives only blink frequency, blink time when there is abnormal fatigue, the integration will be as follows:

(1) input and output variables to establish the membership function: for two input variables closed one eye blink of time and frequency, and an output variable (the drivers fatigue state), three different levels are defined fuzzy sets, for each variable, select the appropriate membership function. Described as follows: blink frequency = {fast, medium, slow}; blink time = {short, medium and long}; fatigue = {not fatigue, mild fatigue, fatigue}. In this article, use inductive reasoning to determine the membership degree method, using triangular membership functions.

(2) fuzzy rules and fuzzy reasoning to establish: Fuzzy is the precise measurement of value has been normalized to the input variables corresponding to the transition of the domain, and then defined by the membership function to translate them into appropriate fuzzy linguistic variables, membership is used for fuzzy reasoning. In this paper, the input variable for the blink of an eye blink of time and frequency, we processed the images collected by the case of closed eyes open, and frequency conversion for the blink of an eye blink frequency of fast, blink frequency, the blinking frequency of slow, long blink, blink time, short blink vague language. Because we blink frequency and blink time for the selection of three levels of fuzzy sets. Therefore, its up to each other was 32 = 9 control rules, reasoning as follows Table 1:

(3) fuzzy clear: the fuzzy definition of is to use fuzzy logic fuzzy variables after the operation are translated into actual amount shown. The paper used centroid method, which is calculated using the following formula. R: Fuzzy controller output; k: number of rules; xi: i-a rule of membership; Fi: s i-rule membership function center of mass value. The higher the fatigue that the more fatigue, eyes closed in this article the importance of time than a blink frequency, blink duration because the longer the representative of the eyes, eyes closed in the blink of an eye during the longer, whether or not the driver fatigue , eyes closed longer representative of the higher risk.

experiment result: We collected 50 drivers in the fatigue section of the video were used alone blink time, blink frequency, and two feature fusion The membership value judgments, experimental results show that the correct rate of fusion was significantly increased. Similarly, when there is yawning, fatigue characteristics lane road excursion, we fused the same way, the system accuracy of 95%.

3 Conclusion

full text of fatigue based on ZigBee technology, a comprehensive warning system to monitor the introduction, and the system design, key technology, the system has unique advantages: (1) using wireless sensor networks based on ZigBee technology, avoiding the high cost of vehicle wiring and interference a serious problem; (2) using a multi-sensor information fusion technology, better analysis and decision to driver fatigue behavior. Therefore, based on ZigBee technology vehicles to reduce driver fatigue monitoring system of traffic accidents due to fatigue caused by a certain significance and better prospects. This innovation: ZigBee network is proposed and implemented multi-sensor fusion with a combination of fatigue monitoring system for vehicle operation and scalability, and effectively improve the fatigue to determine the accuracy and enhance system robustness.