EXPERIMENTS AND RESULTS
EXPERIMENT 1:
Measuring attention Levels
The first experiment conducted was the attention level detection from the mindwave mobile device. In order to implement this, we built a simple LED circuit which consisted of 9 LED's (3 Red, 3 Green, 3 Blue) which were connected in parallel across resistors. The principle of this test was to light up the group of LEDs based on the subjects attention level, the attention level detected from MWM2 ranges from 0 - 100. All three team members took turns to test out the attention levels. We were able to classify the attention level into three states, Low attention with only yellow LEDs turned on, Mid attention with yellow and green LEDs on, and High attention state with all the LEDs simultaneously turned on.
The second experiment entails the measurement of blink strength. To carry out this trail we performed the following steps, the raw EEG signal was detected with the help of the mindwave mobile device. A program was written in Matlab to process the raw EEG data collected from the mindwave mobile device. After processing the data we were left with just the blink strength data. This blink strength data was then given to the Arduino as input and a sketch was written to perform functions based on the blink strength which were in the range of 0 - 255. We tested with two possible ways (i) Lighting up LEDs (ii) movement of the robot. The blink threshold was set to 110. From that three different range was set up to >= 350. A soft blink was programmed to light up three green LED or the right motion of the robot. Similarly, medium blink strength was programmed to light 6 (3 Yellow and 3 Green) LED or backward motion of the robot, when the range of the blink magnitude was between 150 - 250. Anything above 350 was considered a hard blink which was programmed for the forward turn. This system accuracy and reliability had been considerably less.
Due to accuracy issues encountered in the earlier experiments, we decided to include blink count detection along with blink strength ( magnitude ) to perform the maneuver functions of the robot. The original program was modified to accommodate the blink count as a command. Based on the blink count, the necessary functions were modified. First, we added an emergency stop with a forceful hard blink which completely stops the robot( the blink strength will be maximum at 255). The forward motion was calibrated to 1 blink, the backward motion with 2 blinks, the left turn with 3 blinks and right with 4 blinks. The blink count will only start after one blink so that there is no delay or wait time. The blink count we set is for two and a half seconds. So the count is calculated every 21/2 seconds. It will remain in the same state if you are not doing anything till you stop or give any other command to it. With the new functions, it was much more defined and easy to maneuver the robot. We conducted several trials and tested the approach among all three team members and also a fourth subject to see if the calibration was generalized to anybody using the device. If there is a poor signal coming in it will be zero. Blink detection accuracy test: Table below shows the testing of the blink detection accuracy of the Mindwave Mobile. Normal blinks refer to the involuntary or voluntary rapid closing and opening of the eyes. Forced blinks refer to voluntary ones that are hard and forceful. After blink count detection, the accuracy is 100 % for example if the system detects 1 blink it always moves forwards all the time and vice versa.
Measuring attention Levels
The first experiment conducted was the attention level detection from the mindwave mobile device. In order to implement this, we built a simple LED circuit which consisted of 9 LED's (3 Red, 3 Green, 3 Blue) which were connected in parallel across resistors. The principle of this test was to light up the group of LEDs based on the subjects attention level, the attention level detected from MWM2 ranges from 0 - 100. All three team members took turns to test out the attention levels. We were able to classify the attention level into three states, Low attention with only yellow LEDs turned on, Mid attention with yellow and green LEDs on, and High attention state with all the LEDs simultaneously turned on.
EXPERIMENT 2:
Measuring blink strength
Measuring blink strength
The second experiment entails the measurement of blink strength. To carry out this trail we performed the following steps, the raw EEG signal was detected with the help of the mindwave mobile device. A program was written in Matlab to process the raw EEG data collected from the mindwave mobile device. After processing the data we were left with just the blink strength data. This blink strength data was then given to the Arduino as input and a sketch was written to perform functions based on the blink strength which were in the range of 0 - 255. We tested with two possible ways (i) Lighting up LEDs (ii) movement of the robot. The blink threshold was set to 110. From that three different range was set up to >= 350. A soft blink was programmed to light up three green LED or the right motion of the robot. Similarly, medium blink strength was programmed to light 6 (3 Yellow and 3 Green) LED or backward motion of the robot, when the range of the blink magnitude was between 150 - 250. Anything above 350 was considered a hard blink which was programmed for the forward turn. This system accuracy and reliability had been considerably less.
EXPERIMENT 3:
Blink Count
Blink Count
Due to accuracy issues encountered in the earlier experiments, we decided to include blink count detection along with blink strength ( magnitude ) to perform the maneuver functions of the robot. The original program was modified to accommodate the blink count as a command. Based on the blink count, the necessary functions were modified. First, we added an emergency stop with a forceful hard blink which completely stops the robot( the blink strength will be maximum at 255). The forward motion was calibrated to 1 blink, the backward motion with 2 blinks, the left turn with 3 blinks and right with 4 blinks. The blink count will only start after one blink so that there is no delay or wait time. The blink count we set is for two and a half seconds. So the count is calculated every 21/2 seconds. It will remain in the same state if you are not doing anything till you stop or give any other command to it. With the new functions, it was much more defined and easy to maneuver the robot. We conducted several trials and tested the approach among all three team members and also a fourth subject to see if the calibration was generalized to anybody using the device. If there is a poor signal coming in it will be zero. Blink detection accuracy test: Table below shows the testing of the blink detection accuracy of the Mindwave Mobile. Normal blinks refer to the involuntary or voluntary rapid closing and opening of the eyes. Forced blinks refer to voluntary ones that are hard and forceful. After blink count detection, the accuracy is 100 % for example if the system detects 1 blink it always moves forwards all the time and vice versa.
Note: Photos and videos are included in the photo gallery section.