Research involving The Posture-Based Model
During the 2002-03 and 2003-04 academic calendar, including the summers of 2003 and 2004, I worked with professor Jonathan Vaughan and aided him in collecting data for his Posture-Based model.
Abstract (Summer 2004)
Stephanie Godleski '05 and Aram Kudurshian '06
Professor Jonathan Vaughan
In our everyday lives, we carry out various movements quickly and efficiently without giving them much thought. Over the years, scientists have tried to devise models that explain the whys and the hows of our movements. This is a difficult task due to an infinite number of postures and trajectories available when performing various tasks. The posture-based model is one of such attempts to break down motion planning and execution using a computerized computational model.
The model attempts to create lifelike movements by adhering to four assumptions about how movement is planned. Under the first assumption, the goal posture must be chosen from an array of stored postures. The stored postures are then evaluated with respect to a constraint hierarchy for the specific task. The second assumption states that the final posture is found through two steps: a stored posture that comes closest to the goal posture is found, then that posture is modified to create a better posture. Under the third assumption, postures are generated until a dynamic deadline is reached. The process of selecting a goal posture during a specific amount of time is called diffusion 'til a deadline. The fourth assumption provides for a via movement to allow for obstacle avoidance without having any net affect on the final posture.
This summer, we continued to modify and tweak our 3-D apparatus which we developed last summer (in addition to capturing some data), as well as, developed a new apparatus. This new apparatus was simply two pegboards connected together with twelve targets pasted on to it. These targets, in the shape of a bulls-eye contained a small blue circle as well as a larger, black circle - allowing us to measure accuracy. Furthermore, the apparatus was created to allow obstacles, either a horizontal or vertical bar, to require the participant to reach around an obstacle, and thus use a via movement. Besides the physical apparatus, a computer program written in REALbasic was also written to display instructions to the participant which guided them through the targets. The program allowed us to vary the order and length of the experiment as well as provide visual information about where to hit the target.
Data was captured during a four-day stay at Penn State University using the Opto-trak system. By placing multiple IREDs on the participants and the foil that they held, we were able to capture each movement they made to and from the targets, as well as, the trajectory used. During the final week of our research, we began the long and complicated, although rewarding process, of building MATLAB algorithms to display our results digitally. This is the first step in moving towards validating the posture-based model.
Poster (Summer 2004)
Stephanie Godleski '05 and Aram Kudurshian '06
Professor Jonathan Vaughan
Clicking on the thumbnail will open a larger version of the poster.
Poster (Summer 2003)
Brian Burkhart '04, Robin Gane-McCalla '07, Olamide Harrison '05, Catherine Johnston '05, and Aram Kudurshian '06
Professor Jonathan Vaughan
Clicking on any thumbnail will open a larger version of the poster.
