******************************************************************* *** *** *** *** *** DIALY LOGS - NALAKA KAHAWATTE *** *** *** *** December, 2006 *** *** *** ******************************************************************* ******************************************************************* Friday 12/01/2006 2.00 - 5.00 pm Further studied the rUNSWift code using the code itself, Noboyuki's thesis and behavioral control along with the Python tutorial. **** The reason why the runswift bots were continuosly spinning around is that in there code there is a slight head motion delay due to mechanical motor control of the head. The dog potentially looses track of the ball. He keeps searching for the ball spinning around. I remember, even though we kept the ball on the dog's feet, it did not identify the ball. I believe this is because runswift bots use a method of judging the distance to the ball by looking at a point ahead of the ball from top of the ball. For diagrams which describes, please refer to page 30 of Nubuyoki's thesis. Because of this head movement, the chance of missin the ball while spinnin around is greater.. Next time let's try keeping the ball a little bit further to check if they identify. According to their exeperiments, 0.5 m is ideal distance. Something interesting i came across. Thought this would be helpful. The runswift dogs do not have the same routine of searchin the ball. in fact, it has seperate routines which are efficient at different scenarios. according to the situation, the search routine changes. for example, they have different routines, namely.. Ball Hint, Action Hint, Spin Hint, Follow Hint, etc. ( Please refer to page 33 for further clarifications ) *** Since there r no walls for competition this time, if one team kicks the ball out of the ring, notification should be sent to all the players on the field via a wireless message notifying them that the ball has gone out and which team kicked it out. runswift uses another important tactic. by sending this msg, the runswift robots are notified where the ball might be placed. As soon as this msg gets to dogs, they will turn to the predicted spot where the ball is without wasting time looking all around the field.