MR688 Project Sample Test Report File Status Project Name MR688 Date created 2011-12-21 [ ]Draft Document Name MR688 Project Sample Test Report [ ]Release Version 1.1.3 Privilege Level Normal [ ]Revision Pages 5 1. Introduction: In the field of commercial transport, driver fatigue is a very important risk factor. Working long hours, lack of sleeping, night driving, all of these will cause fatigue driving. Fatigue driving caused increase of car accidents. In the past ten years, CareDrive R&D team has been concentrating on technology research for fatigue driving prevention: How to prevent fatigue driving through technical means, so as to reduce the risk of commercial transport, protect driver s life safety. Hao Nai Industry (CareDrive R&D team) work together with Chinese government s driving safety Laboratory, study and research the various characteristics of driver fatigue, as well as the technology of real-time identification for fatigue driving. This project s objective is to estimate the effectiveness and reliability of fatigue driving technology, analyze the influence by reminding and stimulating to driver when the driver is under fatigue driving, commercialize the technology of the fatigue driving prevention. 2. Fatigue driving recognition technology 2.1. Based on the detection method of the driver s physiological index. The drowsy driver s physiological index will deviate from the normal driver s physiological index, then we can judge whether the driver gets into the fatigue driving state through the driver s physiological index. Currently the detection methods includes: the measurements to the driver s EEG, ECG and others. 2.2. The detection method based on driver s physiological response characteristics: Judge the driver s fatigue state through the characteristics of eyes movement and head movement. 2.3 Detection method based on driver s operation behavior Judge the driver s fatigue state through driver s operation behavior such as steering wheel operation. 2.4 Detection based of vehicle s driving trajectory: The driver s fatigue driving state can also be speculated by the vehicle s trajectory change and lane departure information. 1 of 5
3. The detection technology what MR688 used. Caredrive R&D team studied and researched all the related fatigue driving technologies, we found the existing technology is difficult to simultaneously meet all the conditions of non-contact, easy detection, fast response, high reliability and low error. In the research of human eyes response characteristics, CareDrive R&D team got breakthrough discovery: During the process that human get into fatigue state, the pupil will appear the changes which can detected. By identifying the driver s pupil change, the fatigue can be quickly discovered when the driver gets into fatigue state. Except for the research of human eyes response characteristics to the fatigue driving, CareDrive R&D team made deep research on the computer recognition technology of the pupil changing. Therefore, on the product of Driver Fatigue Monitor MR688, we realized face recognition and real-time measurement of the pupil image. CareDrive R&D team also made face feature and pupil characteristics research for different population. Through the way of feature database, we achieved high reliable identification of pupil characteristics. Based on the face recognition and measurement technology, MR688 can also identify the drivers who wear glasses and sunglasses, the drivers looking around, the drivers of different skins (white, black, yellow, brown). Meanwhile, CareDrive R&D team also made research in the recognition and measurement technology under different weather and different lights, all these researches and developments make MR688 can realize to work under all weather, work at daytime and night, suitable for different population. 4. Verification test for MR688 s technology Test 1: Product effectiveness test based on pupil detect technology. Test 2: MR688 Road Test Test 1: Test name: Product effectiveness test based on pupil detect technology. Purpose: Make the effectiveness confirmation of fatigue detection technology which is based on detecting the pupil change (test in the laboratory) Test method: Set 4 groups according to different time (daytime/night) and different activity (monotonous/excited) Select a number of volunteers, using a random manner incorporated into grouping. Let volunteers engage in the activities which are under computer s automatic control (For consistency and automatic statistics). In the process, using emergency response time s measurement to judge whether the volunteers have entered to the drowsiness state or not. In the process, using video camera to automatically record volunteers activities. In the process, using the pupil changing detector(mr688) automatically detects changes of the pupil, and using signal logger to record the alarming information automatically (During the test, MR688 S alarm tone is off, will not alert the volunteers). 2 of 5
Grouping Test: group Test time activities 1 11:00 Reading German grammar textbooks from computer screen 2 23:00 Reading German grammar textbooks from computer screen 3 11:00 Playing the driving game from the computer screen 4 23:00 Playing the driving game from the computer screen Note: Volunteer participants in the test have never learned any German. As most people are easy to doze when they are reading boring text, that s the reason why choosing to read this. Test Process: 1) Making independent test for each volunteer in a room with comfortable temperature. 2) Before starting the test, watch 20 minutes video of Animal World to make volunteers relaxed. 3) Volunteers make 5 minutes exercises of pressing the button when getting the screen prompts.(this is to let volunteers get used to test method, reduce the test deviation caused by not familiar with the operation.) 4) Then begin formal testing activities. The contents for different groups are different, the contents are same for the volunteers in the same group. 5) The test activity lasts two hours, during the activity, computer screen will pop-ups from time to time (give out a pop in an average of 5 minutes), remind volunteers to press the button. 6) After volunteers see the prompts, press the button. 7) Computer automatically records the response time of volunteers. 8) At the same time, the logger automatically records the alarm signal of MR688 during the test process of volunteers. 9) Through the comparison between the saved response time of volunteers and recorded alarm signal time of MR688, verify the effectiveness of MR688 detecting the fatigue state. Driving is an activity that needs focusing and response, it requires drivers are able to make a timely response to the changes of road and environment. The test is through the test of the volunteers response time, judge whether the volunteers get into drowsiness state or not. Determine the sleepy threshold according to response speed>=1s Test records, please see Based on product effectiveness test of pupil detection technology Data Records Test data analysis, please see Based on product effectiveness test of pupil detection technology Data Analysis 3 of 5
Test Results: Note: Judging criterion which drowsiness occurs is: response speed>=1s. test group test persons Persons completed the test Fatigue persons (judge according to the response speed) MR688 Alarm records 1 15 15 10 9 2 15 15 13 13 3 15 15 0 0 4 15 15 6 6 The conclusion of tests: 1) One volunteer s response speed drop lower, it wasn t detected by MR688. From the video, it displays this volunteer can read the text normally after ending the test. 2) All the volunteers who have alarm records, their response speed is more than one second. 3) The detection effectiveness rate is: 28/29=96.50% Mis-judgment rate: 0% Test 2: Test name: MR688 Road Test Purpose: Through road test, further verify the driver fatigue monitoring effectiveness of MR688. Test method: Select three trucks from the logistics company to make verification test. Every truck has two drivers alternately, it runs 12 hours on the road everyday, continue 7 days test. Trucks are installed with MR688 (turn off the alarm tone), also installed with video camera and automatic signal logger. When the truck s speed exceeds 10KM/H, the signal logger automatically starts recording the signal output by MR688, and make video recording at the same time. When the Monitor (MR688) detects the alarms, it does not alarm, just automatically send the signals to the automatic signal logger for recording. After the truck returned to the original station, collect the recorded data. After finishing the test, make data analysis all together. When making the data analysis, check the driver s records of the MR688 s alarms, determine the driver s state at that 4 of 5
time. Making the classification statistics for the alarm state. Effective alarm: When alarm is detected, drivers are driving on the road, mental state is not ideal. False alarm: When alarm is detected, drivers are driving on the road, mental state is normal. Data of test result Recorded alarm times Explanation Effective alarm 12 MR688 detects the fatigue in the test, it does not warn the driver (tuned off the alarm tone). After merging the alarms of the same period, the actual alarms are four times, three of them are at night driving. False alarm 4 For 2 times, the driver is possibly not concentrating on driving or false recognition, can t judge from the video. For the other 2 times, MR688 mis-judged due to the driver s movements. Test Conclusion 1) The effective alarms are four times, which indicates that professional drivers drive at night or drive long distance, they will appear fatigue and distraction. 2) Driver Fatigue Monitor MR688 can effectively detect driver fatigue state, can warn the drowsy driver and enhance the safety of driving. 5 of 5