|
| Automatic Measurement of Behavior Using Inertial Measurement Units |
| Sunday, May 27, 2018 |
| 10:00 AM–10:50 AM |
| Manchester Grand Hyatt, Seaport Ballroom A |
| Area: PRA; Domain: Applied Research |
| Chair: Nathan Blenkush (Judge Rotenberg Educational Center) |
| Discussant: Erich K. Grommet (Troy University) |
| CE Instructor: Nathan Blenkush, Ph.D. |
| Abstract: We assessed a novel method to measure behavior frequency. Utilizing the Life Performance Research's Inertial Measurement Unit (IMU) device and a data filter, we counted response frequency derived from Euler angle and linear acceleration data associated with instances of simulated self-injurious behavior. When compared to a frequency counts of trained observers, the response rates were equivalent. Visual inspection of graphs showing Euler angle, linear acceleration, and temporal location (determined by observers and the data filter) show the filter is accurate when compared to observers. The measurement procedure successfully differentiated between hits to the head and leg, as well as less forceful touches to those same areas. Taken together, the data show the feasibility of automatic measurement of behavior with the potential to improve aspects of research and practice in behavior analysis. |
| Instruction Level: Intermediate |
| Keyword(s): inertial measurement, measurement |
| Target Audience: Board Certified Behavior Analysts, |
| |
Automatic Measurement of Behavior Using Inertial Measurement Units |
| NATHAN BLENKUSH (Judge Rotenberg Educational Center), Gary Woo (Judge Rotenberg Educational Center), Robert W. Worsham (Self Employed), Nick Lowther (Judge Rotenberg Educational Center), Jason Coderre (Judge Rotenberg Educational Center), Tristan Webbe (Judge Rotenberg Educational Center), Asli Unver (Judge Rotenberg Educational Center) |
| Abstract: We assessed a novel method to measure behavior frequency. Utilizing the Life Performance Research's Inertial Measurement Unit (IMU) device and a data filter, we counted response frequency derived from Euler angle and linear acceleration data associated with instances of simulated self-injurious behavior. When compared to a frequency counts of trained observers, the response rates were equivalent. Visual inspection of graphs showing Euler angle, linear acceleration, and temporal location (determined by observers and the data filter) show the filter is accurate when compared to observers. The measurement procedure successfully differentiated between hits to the head and leg, as well as less forceful touches to those same areas. Taken together, the data show the feasibility of automatic measurement of behavior with the potential to improve aspects of research and practice in behavior analysis. |
| |
Developing IMU Filters and Measuring Behavior in Applied Settings |
| DYLAN PALMER (Judge Rotenberg Educational Center; Simmons College), Joseph Tacosik (Judge Rotenberg Educational Center) |
| Abstract: Developing Filters Using data generated from inertial measurement (IMU) units is a difficult, time consuming, and essential exercise. Here, we summarize the steps taken in developing a filter for one self-injurious response. We review raw IMU data and illustrate the process of identifying threshold values for counting. Once an IMU filter is created, there are a number of practical problems that must be solved. Battery life, Bluetooth connectivity, data storage, IMU placement, and many other variables affect the utility of measurement. Finally, we summarize our experiences with artificial intelligence filter development and various inertial measurement devices. |
|
| |