Two dimensional trajectory forecast training set 'TraSe' as a industrial control task imitator
2008
Mednis, M., Latvia Univ. of Agriculture, Jelgava (Latvia). Faculty of Information Technology. Dept. of Computer Systems. Biosystems Group | Vilums, S., Latvia Univ. of Agriculture, Jelgava (Latvia). Faculty of Information Technology. Dept. of Computer Systems. Biosystems Group | Stalidzans, E., Latvia Univ. of Agriculture, Jelgava (Latvia). Faculty of Information Technology. Dept. of Computer Systems. Biosystems Group
A training set 'TraSe' has been developed for university students specialising in information and communication technologies for industrial applications to simulate physically two types of industrial computer control tasks: 1) early deviation recognition and 2) real-time control. The training set addresses some methodological issues that can be solved by training sets based on real physical processes. Two classes of computer control tasks can't be simulated by a computer as random number generators can't well replace real physical processes in their wide variety as computer generated perturbations of the system are less creative and easier adjustable compared to the ones of a physical process. 'TraSe' use the feature of a rolling ball to keep predictable direction on a horizontal surface. 'TraSe' is developed as a game where two players develop programs and algorithms allowing computers to compete for better performance. Defending player using starter mechanism has to hit by his ball the ball started by attacking player. Computer of defending player can see actions of attacking ball by a web cam. In case of a single player following exercises can be performed: determination of trough angle, forecast of the ball trajectory by speed and direction, determination of the table angle by the trajectory of the ball, avoiding visible obstacles on the way of the ball, scanning of invisible obstacles on the table by registering perturbations of the ball trajectory. Tasks can be solved using methods of artificial intelligence, statistics and other methods. Time and information constraints request creative development of algorithms. Thus skills of industrial control task performance optimising can be developed. The training set and different samples of training tasks for one and two players are described.
Показать больше [+] Меньше [-]