2009 Volume 21 Issue 3 Pages 110-117
We have introduced a system which detects the action of getting out of bed using an ultrasonic array sensor to monitor head position. In actuality, the sensor on the ceiling cannot be always located just above the center of a bed. So, we have to evaluate how our system can be robust when the sensor cannot be located on the ceiling just above the center of a bed. Also unwanted ultrasonic derived from the next sensor may disturb our system. Then a new test environment was created to be able to change the location of the sensor compared to the bed by moving the bed back and forth, to the right or to the left 20cm each direction. Also another sensor located 210cm away on the ceiling was added to the system. In the two situations, 7 subjects got out of bed 15 different ways. Two series of test have been done. First, we tested using a PC to execute the program of the getting up from bed. The data was collected every 200ms. Second, we tested using our new prototype, collecting the data every 100ms. We thought the shorter interval of would keep up the higher detection rate. In conclusion, we estimate the correct detection rate as follows in the test environment:
•93.3% with an error rate of 1.9% and a loss rate of 4.8% for sitting up in the bed.
•93.3% with an error rate of 0.5% and a loss rate is 6.2% for getting out of the bed
We could not achieve better detection rates due to the circumstances, which are more critical than the previous report. We think we could better evaluate the robustness of detection algorithm in an actual environment. We think we should enhance our algorithm and especially lower the loss rate in the future.
