On Wednesday we found out that the stepper motor and DC motor, connection’s were not working as intended. We decided it would be a good idea to replace all of the wires as well since everything was quite messy, the rest of the time we used to develop a solution that would solve the problem that the DC motor would slow down when connected together with the stepper motor using the same PSU. We tried different circuits and solutions including making a battery slot and using a 12V battery for the dc motor then running the stepper motor separately. Finally at the end of the day we changed the driver and that seems to be working OK for the DC motor but not for the stepper. We should be able to fix it this week.

Fixed the problem of the DC motor not receiving enough power by swapping out the stepper driver.

This week after more testing we noticed there was difficulty to recognize between the queen and black ten playing cards, as the way we programmed our image recognition we only got the 0 from the 10, and similar for the Q from the Queen. This was especially true for the red playing cards as they have less of a contrast to the white background, we therefore saw a worse outline of the numbers. This happens because we find the playing card number with the raspberry pi camera image and draw a square around the number. Then we zoom in on the square so that we get a picture of only the playing card number and compare that to a saved image of different playing card number. We adjusted the mask color bounds to make the red cards clearer. Also, by changing the size of the square generated around the contour of the number, we managed to collect the 1 and 0 from the 10, which allowed the image recognition to easily see the difference between the different pictures that we take and compare, without affecting the results of the other cards that were already easy to recognize. The square is increased in a certain percentage, 60% to the right. So further away playing cards with smaller squares are increased in the same percentage and not a static pixel increase, that would be a big increase in some cases and too small of an increase in other cases. Our image recognition is however still sensitive to light especially red, and we therefore continue to test different lights and lighting conditions. Since we need light to remove shadow’s in the images and we can’t use too bright light, as that will create bright spots on the image. We also continued to test and fine tune tests we did last week, like the function that adjust parts of the program based on how many cards has been dispensed. This is needed since the more cards are dispensed, the farther away the cards get from the camera and to still find the playing card number and not the other contours (other black parts) in the image, this is needed.

The VGA connecter was soldered so it could be connected to breadboard.

VGA port connector, to allow connection of VGA cable to VGA connecter.

This post was written by Seonhee Kim