Customer Solutions

Developing a Dynamic Visuomotor Inspection Task Simulation System with LabVIEW and NI Vision

Author(s):

Praveen J. Chellakumar, National Institute for Occupational Safety and Health (NIOSH), Engineering and Control Technology Branch (ECTB); Kiran Kumar V. Kunderu, NIOSH, ECTB; Aaron W. Schopper, NIOSH, ECTB

Industry:

Research

Product:

Data Acquisition, LabVIEW, Vision

The Challenge:

Developing an accurate, cost-effective simulation system to support a research program that characterizes and studies human visuomotor inspection activities.

The Solution:

Creating an integrated, server-based, menu-driven, graphic simulation system using the NI Vision Development Module, LabVIEW, and NI DAQ hardware that provides the capability to (a) vary the nature and intensity of the perceptual and motor demands associated with the tasks, and (b) automatically capture associated human performance and physiological data for eight participants simultaneously.

Frame Creation VI

Simulating Upper Extremity Tasks
The Health Effects Laboratory Division of the National Institute for Occupational Safety and Health has initiated a research program to better understand and characterize work performed with the upper extremities – the operators’ hands, arms, and shoulders. To effectively and efficiently pursue this research program, we needed to develop an integrated, server-based, dynamic simulation of an upper extremity task that is capable of presenting the simulated tasks on eight monitors simultaneously while collecting tightly coupled data from multiple physiological and performance-related sensors that are either worn by the participants or used by them to perform the tasks. We used National Instruments LabVIEW for our project development for its ease of use and NI data acquisition, imaging, and analysis tools.

Building the Simulation System
We built the system on a client-server architecture using a local area network (LAN). The server was a dual processor workstation running Windows 2000, and the clients were low-end desktop computers running Windows 2000 with 17 in. XGA color touch-screen monitors. The server sent the video signals required for the extensive, dynamic simulation to all clients via a video splitter. We captured participant performance and physiological responses with an NI LabVIEW program. The server software had three modules – frame creation, frame movement, and data analysis.

With the user interface, the researcher could select from a variety of simulation options: background color, object color, flaw color, object density, flaw density, object velocity, and number of frames. The researcher also could create unique frames composed of target and nontarget objects of varying difficulty. We used NI Vision VIs to generate circular, nontarget objects of different diameters. The overlaying flaws of varying size and shape at different locations generated the target objects. We then randomly filled the interior surface of each object with small particles of varying complexity and orientation to increase the visuoperceptual demand. Using a complex uniform – but random – distribution algorithm, we overlaid these objects on the frames such that all objects did not appear to be aligned in horizontal rows. We saved the frames on the server as image files and the frame parameters as data files to support frame movement and data analysis respectively.

We extensively used image acquisition window VIs running on the server for frame movement to simulate objects moving from left to right across the screen. We achieved uniform object velocity through hardware timing using a digital counter on the PCI-6035E. Then we used two photo sensors, connected to either end of the monitor at the same horizontal level using fiber-optic cables, to validate the uniform object velocity. The photo sensor circuits differentiated between dark background (“low”) and the objects (“high”) in the frame by measuring the low-to-high and high-to-low transitions, which helped us calculate the time difference between the object entry and exit from the display. Then we conducted a series of repetitive tests to validate the uniform movement for different object velocities and saved the test start time to the server as a data file for use during data analysis.

Participants identified target objects by tapping the monitor with a pointing device. The pointing device was specially designed to record both accelerations and finger grip forces. We used a PCI-6023E to acquire pulse-width modulated output from the Micro Electro-Mechanical Systems (MEMS) accelerometer and the finger grip-force measurements (analog inputs) from a customized pressure sensor. We time-stamped the participants’ responses and saved them to a file for further data analysis. To reduce human resources, we time-synchronized the LabVIEW programs on clients and remotely started/stopped them using a VI server.

We recorded participant heart-rate measurements using Polar belts and receivers – the latter were mounted at each client station – and used the eight NI timing I/O counters on the PCI-6602 for measuring the R-R intervals in milliseconds for heart-rate variability (HRV) analysis.

Analyzing the Data
We used the analysis VI, a computation-intensive program, to process the data from the client computers. When the researcher specified which client’s data he needed to analyze, the program traversed the folder subdirectory tree structure and automatically linked all the files (test start time, frame parameters, and participant responses) needed for data processing. Using the time information, an algorithm determined the frame in which a particular response was made. Using the X-Y coordinates of the response, the program determined if the participant responded on a target object, nontarget object, or a blank space. The program also tracked the accuracy of the participant’s responses as a function of the types of flaws (for example, size, location, and background densities).We saved these results as a data file on the server for subsequent statistical analyses.

Rather than taking the more typical approach of studying each participant separately, researchers can use the flexible, integrated, dynamic simulation system we successfully created to reduce multistudy research program data collection and analysis time and effort by a factor of eight. With its automatic data reduction capabilities, the system not only yields accurate data, but it also markedly reduces the human resources that would otherwise be required.

For more information, contact:
Praveen J. Chellakumar
Computer Engineer
National Institute for Occupational Safety and Health
Engineering and Control Technology Branch
Morgantown, WV 26505
Phone: (304) 285-6168
Fax: (304) 285-6265
E-Mail: praveen.chellakumar@cdc.hhs.gov
Web: www.cdc.gov/niosh/