Concern
ECE 4760
Upon the first few simulations and tests, we noticed a significant problem. Mechanically, the structure was shaky and rattled a little bit when the gears turned. This was because we did not fit the acrylic with a ball bearing because the price of ball bearings on McMaster for the size we needed were a bit too high. So, we simply drilled a hole and put an axle through. The reason the wheels were a bit shaky was a result of the rod being slightly smaller than the hole, which causes angling of the axle and added resistance to the torque motion of the wheel.
While we used it to remedy an issue before, another possible problem was the front wheel. The front wheel was cut out of acrylic using the laser cutter. This implies that the wheel is actually not symmetric. When cutting, the laser is pointing straight down and melting through, but the residual acrylic will actually collect towards the bottom while it is still in the machine. This causes for a slight unevenness which results in different resistances when turning in different directions.
A third concern is regarding the precision of the gears and the external environment. Ideally, the gear would act as a perfect wheel with a nice traction so that it would turn a perfect amount every time. Unfortunately, not only is the gear non-ideal, the environment is as well, which causes the gears to “slip-n-slide” occasionally when turning. There were also times when only one gear would drive the car while the other gear was driving in place. Problems such as this one probably could have been resolved with a better wheel system.
The robot often takes very long to execute all the drawing instructions in part because we set motor speed to low on purpose for stability in mechanical parts. This had made this project impossible to demo full execution of 5 color drawing.
Future Improvement
Result and Concerns
Feedback System:
A proper feedback system would have improved the accuracy of our machine. The idea behind a feedback system would be to gather input from the external environment and tell the car what to do next based on this reading. Since our car did not always go perfectly straight, there was a good chance that after every line increment, the car would slowly begin to deviate from the path and these errors would sum up as the program ran for longer. A feedback system would have prevented, or at the very least, alleviated such things since the car would then have an idea of how to correct itself as it was printing instructions.
Wireless Transmission:
As a result of the bulky serial port cable, the car was occasionally dragged to cause some sliding when turning. It would be best if we could eliminate any wires or cables that could have been transmitted through air instead of a physical cable, but we had enough trouble attempting the optical mouse feedback which resulted in a time crunch. Otherwise, this would have been a good idea and might have fixed a few errors regarding cables from before.
Movement System:
We are currently relying on two stepper motors. However there are implementation that allows very flexible and a system with large freedom of movement by using balls and drivers. By using 3 balls that are 60 degrees, we can have any kind of motion using vector calculation of three different motion.
Multiple pen:
Instead of using one pen if we had 5 pens, we are guaranteed to have 5 parallel lines. This would minimize number of turns the robot have to make but this design is not a solution but mere improvement to the performance.
A More Robust Hardware Design:
Due to the fact that the car wobbled a bit, possibly causing deviation from the real path, it would make more sense to make the hardware such that it was more stable when traveling. This could be done by using actual wheels instead of gears, refining the third wheel, sockets for all loose parts such as the MCU, the battery, and the breadboard. Most importantly, ball bearings in place of a hole would probably have reduced the wobbling quite a bit when shaking it.
Flexible Software:
Current design of software DOES NOT only any circular drawing or arc-like drawing. For more practical purpose, we should start considering more flexible movement and no more dot matrix printing.
