Project week 3

Saturday, Sunday, Monday, Weensday (08/01, 9/01, 10/01, 12/01)

From 10 o’clock until 17:00

Participants: Michael Kaalund and Brian Hansen

Shcedule of the week
- Get the robot to define outline
- Search drive with and without RGB sensing
- Find atleast one tower.
- Send koordinate for tower
- construct behavior diagram

Progress (notes)
- Constructed flowchart for the different states
- We had to implement a wall following function because the direction of th NXT was wery unprecise
- The simple wall following was not enough so a PD regulator was implemented
- The system is designed to get a faraway destination coordinate and by knowing its own coordinate it define all the necesary coordinates on the rute with a grid of 10 cm.
- for every 10 cm it corrects its coordinate drive if the wall is near and it makes a search sweep for the reed tower.
- Debug PC connection used for wall distance measurements and applied coordinate corrections
- Tacho precision for the sensor platform is wery unprecise and needs to be manually calibrated for every try. The problem is that the direction does not return to the original position when a sweep is preformed. The motor control does not register the error it is only visual seen.
- Encountered PD error. After some time the PD results go "beserk". Only solution now is to makes an average of 10 sonic sence at every wall correction. This is time consuming !!!
- Problem if the start position is to farr from wall then it will turn ex. 45 degreess resulting in a colision with the wall.
- Still problems but now the device turn 45 degrees away from wall when it gets to close, but works perfektly when to far away.
- Trying to solve problems by inserting statement that decrease the drive forward(10cm) to two steps of 5cm if the calculated error of the PD is large.
- sweep platform issue seems to be fixed by using rotate_to instead of rotate. Rotate_to have been used before but was buggy meaning that sometimes in turn in the wrong direction.
- Because we use the leJos goto we have not direct control of the motor power that migth be the reason why the motor connected to the C port behaves badly by not stopping at the same time as A or giving the same amount of power.
- an idea of moving the weigth back to try to stop the jumping when it stops doenst seem to work. Changing motor.flt does not help with the problem either
- an idea of the startvalue distance calculation is that the rotation of the platform has not completed before we calculated the distance. therefor we tried to use the function is_rotating to make sure it has completed before distance measurement is started. Even tho it is documented it is not implemented, so we implemented it on over own. the result was that this was not the source of the error.
- after some trial and error the wall following is improved by decreasing driving speed and turning speed.
- implemented function that makes a decision on witch way to turn when it gets to a corner. if left and rigth are blocked it will stop and write game over to the consol
- problem at corners with rotation speed/power, cannot turn enough, and the garbage collector seems to be used often, that migth be because of the constant bluetooth connection.
- tower detection does not work, and no search drive only wall follower and corner detection.

Next time 13/01
- Make precentation program for sonic-RGB distance sence with debug mode
- Make precentation program for Tacho drive to show coordinate errors
- Get the corner detection and drive to work.
- precentation at 15:00