DTSF Conclusion

The last week of the DTSF program when pretty well, I ended up using the QR codes and the mission pads to navigate the drone and update its navigation in route.

I decided to create two different navigation paths, one in which the drone simply takes off reads the QR code, adjusts itself if it’s not in frame and follows the given navigation, then uses a search grid at the end to complete the route and drop off the package. The second navigation route was longer to the point where it actually would navigate the drone down the hall and around the corner. This one was a little tough in that there was a lot of trouble shooting. The difficulties that I encountered are as follows:

  • Weight distribution
  • Height measurement
  • QR code reading

I mention weight distribution because of the hook that I attached to the drone. The hook was taped to the drone via double sided sticky tape. It did not weigh too much but if I taped the hook slightly more to the right versus the left then the drone’s navigation would be impacted by this—shifting it more to the left or the right depending on how I taped the hook on.

Secondly, the drone’s measurement of height. This greatly impacted the reading of the QR code because the drone height measurement was not accurate in that it would return a different height measurement after each launching of the program. Why this is, I’m not sure but it turns out that the mission pads height detection was way more efficient, meaning that if I told the drone to go to a height of 50 it would do so, there would be no inaccurate height measurements.

Thirdly, the QR code reading. This was probably the most difficult thing I had to do throughout the whole project. The QR code reading method itself was not hard to code but the adjusting of the drone to get the drone to capture a good frame of the QR code, was. But looking back, the main reason why I was having so much trouble was because of the inaccurate height measurements. I’m just glad that we were able to sort out that problem in the end. Additionally, the end goal was to glue a QR code that contained the navigation on a delivery box. However, I previously printed out a picture that was relatively large, so the excess paper created issues for the drone in flight. So, I then tested out different QR code sizes and was able to find the right size that wasn’t too big or too small, and one that did not interfere with the drone’s navigation.

In conclusion, I was able to solve all three of these issues, except for some occasional errors, and I learned that drone programming is far from precise data and straightforward code. A lot of the time you will run into unexpected errors that might drive you crazy only for you to realize that it was the slight covering of one of the sensors, or you taped the hook a little too far to the right. I learned how to 3D print and how you really need to carefully think out and measure whatever mechanism you’re trying to print out or else you’ll be rudely surprised after a 3 hour wait when you discover that whatever you printed out is too big or too small. Essentially, my takeaway is that the tedious work is often necessary in order to execute whatever you’ve set out to do. In the future I think I would have tried to finish earlier so I could’ve looked into to voice navigation, in addition to creating more complicated navigation routes. But ultimately, I’m happy with how my project turned out.

Week 5

This week I made significant progress on navigation. Meaning that I’m 90 percent done with that aspect of my project. I was able to get the drone to read the mission pads and detect them and I also implemented the QR codes. In that, now the drone can follow the commands of the mission pads and also follow the commands of the QR code. What was really helpful in speeding up the navigation process was marking with tape where the drone was, and was not able to read the QR code, and also marking where the drone landed at specific distances and seeing if it could accurately land in the same place every time. Next, I simply need to implement the code that Eric is writing that will allow my drone to use a search grid to find the given mission pad. Then, I will finally, (hopefully near the end of week 6) be able to implement code that will enable my drone to put down objects.

I wasn’t able to upload the full video, but the following shows the drone taking off, detecting the first pad, following the first pads directions, then detecting the second pad, rotating, and finally adjusting itself to read the QR code. Then what you won’t see is it eventually adjusting itself correctly to read the QR code, following those directions to reach the third pad which it detects. Then following that pads commands and landing after having executed them.

Week 6 I will be working on the putting down aspect of my project and implementing the grid search for the mission pads.

Week 4

This week I began to think about orienting the camera so that it code detect the qr code downward instead of straight ahead. Eric was able to find a way to do that without having to dismantle the actual drone in order to physically shift the camera. So, I used a pre-made 3D model of a mirror that I attached to my drone that mirrored the camera downward. However, in terms of reading the qr code, that only worked once and the rest of the times it failed to detect and decode the qr code meaning that the mirror must have contributed to some sort of image distortion. Then I decided to 3D print a stand that would hold the drone in place so that it would be easier to get the drone started on its mission. The 3D printing took almost 2 days so I will have to see at the beginning of week 5 if it will work. The idea is that I won’t have to hold it up every time, and waist time trying to orient it in a way so that it can read the qr code, it’ll just be placed on the stand in a fixed position. Furthermore, I’m working on modifying my code so that it can decode both words like “left” and “right and also numbers like “20 and 30” so that it knows how far to go in millimeters instead of my putting fixed numbers directly into my code, which I got to work.

Week 3

I was able to figure out the qr code business. It no longer is adding random commands and is following my directions. Then I figured out how far the drone actually goes in terms of millimeters and if its precise each time it goes a certain distance. The answer to that is it does not precisely move or land in the same spot each time it moves a certain distance. Therefore figuring out how to orient it will be difficult. Additionally I tried and failed many times to print out my 3D model via the resin printer. It simply does not work so I decided at the end of the week to just make the mechanism bigger in scale so that the 3D printer has an easier time of printing it out. I just hope that when it comes to taking out the supports it doesn’t break. Next week, I hope to make substantial progress with the orienting so I can move on to picking up and putting down the boxes using my 3D printed model.

Week 2

This week I was able to make a lot of progress on the QR code aspect of my drone. I used QR API for the detecting and decoding, then created my own method for the drone which iterates through the decoded message containing directions and then assigns each word to drone API code for navigation. It was a struggle in the beginning only because my method wasn’t iterating properly but I was able to find the solution. Additionally, I printed out many QR codes to test out my drone’s decoding and navigation abilities which began to work, but then I ran into a calibration issue. So, I had to go on the tello app and re-calibrate the drone which took a second. I also learned that the drone wings are marked so you must make sure that the right wing is on the right part of the drone. The problem for next week that I will need to fix, is the drone’s navigation inconsistency. Meaning, it will follow maybe the first command after reading the QR code, but it seems to be adding in another command, then skipping the second one and finishing with the third. I’m confused about what’s going on so I’m going to be fixing that next week. Lastly, I began the 3D printing process to print out what the drone will be using to pick up boxes and put them down. However, I did not anticipate that I would be facing difficulty with this aspect of my project. The issue in the beginning at least, was that my design was flawed, so it was interfering with the propellors mobility. Now I’m having trouble with printing out my design, in that the 3D printer is not able to fully print out one of the crucial aspects of the design that will act as a hinge—-another thing I will be working on next week.

Failed 3D Attempts

Video: When I put the wrong propellors in the wrong places

Week 1

I was able to get the basic navigation down for the tello drone, meaning that I gave it commands using the tello drone API. Then I moved on to getting the drone to take pictures and read those images. It didn’t work at first because it looks like you have to install the earlier opencv version because the cv2 gives you errors with the latest version of the opencv. After figuring that out I was able to get the pictures working but my program would fall into an infinite loop each time I ran it, and then I would have to manually stop the program. After working on it with Akriti I realized that it was falling into an infinite loop because I wasn’t actually setting the while loop to false after it entered the loop. After doing so, it worked. However, the images weren’t saving to any folder. I will tackle that problem this coming week then move on to programming the dron to read QR code images.

Week 1: Day 1

Mission Statement: My goal is to modify/program a drone that in some ways replicates drone delivery in that it should be able to identify an object, pick it up, and place it in specified places using a gps/camera. By the end of the program, I hope to have accomplished all aspects of my project in addition to an extra challenge of adding voice control to my drone. However, the voice control will be the bonus part of my project, provided that the standard aspects of my project are done, and I have enough time to work on voice control.

Will be using python instead of java so it will be challenging at first to learn a new coding language, and I will need to practice flying a drone.

The Key components:

  • Navigation: moving the drone from point A to point B– potentially using a gps
  • Pick up/put down: QR code to center it, using an electromagnet to pick it up and release it

Extra Challenge:

  • Voice control: voice commands such as start, and stop.

Goal:

To have everything fully functional i.e. the navigation, the pick-up/ put down, and a basic level of voice control.

More Realistic Goal:

To have at least the navigation from point A to B and the pick up motion down.