Oct 10-16:
During this week , I mainly assisted the team in constructing our first prototype. We chose to build it from foam-core, so I cut out some of the shapes needed to build the different parts of the device (such as the arms and sides). After constructing the basic skeleton, I wrote an Arduino program to control the device for preliminary tests. This involved connecting some buttons to the Arduino on a breadboard, and writing code to generate the necessary PWM signal to run the brushless motor. We had two sets of buttons. One button was used to start the motor, and three buttons on the bottom of the device were used to stop the device when it hit the ground. After I tested this worked on a breadboard, I then soldered wires to the buttons and with the help of the team, attached these and the remaining electronics onto the housing. We then ran some tests as a team indoors (as it was raining) to see if the motor was able to move the device, and it was (explained in the team status reports with videos). I also began thinking of how to create the vectorization algorithm for the control system, and with the help of Lahari and Vikram in the realm of matrices, we landed on an approach that I can begin to implement, keeping me on track,
Oct 17-23:
This week was mainly focused on testing the propulsion of the device outdoors, and seeing what modifications we needed to make. We worked as a team to figure out how to mount the parachutes to the device, and went out the Pausch bridge to drop it. Lahari and I dropped the device from the bridge, while Vikram was at the bottom in order to catch it. The results and consequences of our findings will be discussed in the team status report. Due to the rain, it was difficult for us to do more testing outdoors. At this point, we also found that the buttons placed on the bottom of the device to stop it were not ideal as they could sometimes not be pressed (if the device fell on its side, or on grass), so we decided to move to an ultrasonic sensor that would stop once the device was close enough to the ground. We have also moved to a camera approach for our perception (discussed in group status report), so I helped attach the ultrasonic sensor and camera to the bottom of the device, and cut a slit to allow the wires through. Lastly, as we needed to test the camera on our Raspberry Pi, we used an OpenCV example script to run the Pi Camera, and I helped integrate Lahari’s circle detection algorithm (which we will use to find our target) onto to the existing python script. We found that the script runs fairly well, but more testing needs to be done to determine the true latency of the script. The results from this script will be fed into the vectorization algorithm that I’m working on.