| |
  |
| Predicting Mouse Motion With Genetic Algorithms |
|
 |
|
posted by Editor on Wednesday October 17, @02:35PM
|
|
|
|
|
Pointing and clicking remain the predominant actions in most current user interface designs, but it is usually overlooked how much of a burden these tasks actually impose on the user. The cursor often overshoots the intended target, forcing the user to make a corrective movement in order to reach the desired function. It would be useful if the computer could assist in this process, but how can it do so if it has no knowledge of the user's intent? This project is trying to use genetic algorithms to predict a user's mouse motion based on analyzing information derived from the beginning of the movement. The algorithms works by generating a number of predictors that make guesses about the next endpoint of the user's mouse movement. Based on their accuracy, some predictors are assigned a good or bad reputation. The predictor with the worst record of targeted predictions is removed from the population, and is replaced by a fresh predictor. Over time, a stable population of good target predictors evolves, which is supposed to come up with a pretty good approximation of the real position. The project is not giving out any code, but is asking visitors to participate in a target acquisition experiment (Shockwave) in which you click on a moving target 100 times, from which empirical data can be collected. However, there is also an animated demo (Shockwave) that nicely illustrates the algorithm at work while you move the mouse around.
|
|
|
|
< The Ideal Search Interface | Building Communities In A Virtual World > | |
|
|
|
Don't have an account yet? Go Create One. A user account will let you customize the site's content according to your preferences. It will also allow you to moderate the comments of other users.
|
|
|
|
|
