Robot Photo 25/Feb/2022
The semi-assembled robot as of Friday, February 25th
Robot Photo 4/Mar/2022
The assembled robot as of March 4th, now with a new chassis and complete breadboards, most importantly with a working photoresistor circuit on the front.
Our wheels had little traction on the competition boards, so we added rubber bands around them to try to increase traction.
The AA battery pack was attached to the back of the robot to try to increase the amount of mass on the wheels+motors to hopefully increase traction even further. The wooden block beneath the batteries acts as a spacer to keep the motors from cambering.
This breadboard controls the LEDs that let us see calibration as well as provide light to the photoresistors. The LED on this breadboard blinks to let us know when it is calibrating. There are additionally two resistors that lead to wires that snake below the chassis and attach to the light shield. There, we have LEDs placed to provide controlled photoresistor light. Breadboards are currently attached via their sticky part on the back, though this will likely change as they tend to fall off.
Yes, the light shield is made with a PopTart box (hence the robot's name, Tart). We felt it was easier to make this part with tape and cardboard as it would allow us to easily change its size during development to determine the correct dimensions that would block a sufficient amount of light without disrupting the forward movement of the bot. We plan to keep this as final for now (we like Tart's name and the PopTart logo is fitting).
Here is the 9V battery that powers the Arduino. We chose to have it put underneath the Arduino as this allows us to have more room in the back for the large AA battery pack as well as giving us room to have a spacer to control motor camber. It will likely be more secured to the chassis using tape at some later point.
Here is the Arduino board itself with the (soldered) motor shield attached. We have placed it on the top of the robot as this allows us to easily access it via all 3 breadboards as well as with the USB-Type B cable that attaches to our laptops to let us see the serial monitor readouts. It is held in place with a previously removed bit of the 3D print (clarification under Component Photos -> 01 and 02 (Chassis Schematic and Printed Chassis)
Hidden from this isometric view is the potentiometer circuit, also made on a breadboard. It is simple, each potentiometer being connected to power and ground with an attached jumper wire to take a voltage reading. This reading is translated in the code to alter either speed, proportion, integration, or derivative, as is standard with a PID controller. Breadboards are currently attached with the sticky part on their backs, though this will likely change as they tend to fall off the bot.
Here, we have the main sensor for the robot, the photoresistor array. As with all other robots, it reads the difference between the reflected light, associating the different amount of reflection with either black or white. Using this, we can determine the position of the robot along a line and correct if the bot starts to stray from it. It is connected as is stated in the Lab Manual and has the light shield attached directly to it to block as much light as possible. Jumper wires are taped together to make them look neater. Breadboards are currently attached via their sticky part on the back, though this will likely change as they tend to fall off.
Interactive 4/Mar/2022 Photo
Interact with this photo to see our internal design choices (hover over the yellow boxes to see our insights).
Robot Video 4/Mar/2022
The robot as of March 4th, running in a loop.
Robot Update
9/Mar/2022 - 10/Mar/2022
No photo sadly
We had a minor fiasco with our robot on Wednesday (March 9). One photoresistor's leg broke in half so we had to replace it, but the robot went haywire after that, suddenly and consistently sweeping itself off the track despite going in a straight line for multiple feet. Many hours later, we decided to just replace all photoresistors and eventually discovered that some photoresistors were reading critically low error values no matter where they were placed in the array, so they were replaced with ones that could better distinguish white from black. However, we now had altered the robot so much that we lost about a week's worth of fine tuning.
On Thursday (March 10), we met again and disabled the printing features in the code, which surprisingly made a huge positive difference, likely speeding up the Arduino and letting us get more readings per second with the photoresistors. We also cut the light shield and did some last minute fine tunings to get it as close to what we had before the fiasco as the competition was the next day.
The rubber bands are gone! Instead of using rubber bands to keep the motors firmly attached to the chassis, we super glued them down. We also removed the band that was bringing the motors together horizontally and just stuck with a modified spacer (two wooden blocks on either side of the AA battery pack, all taped together). The rubber bands on the wheels stayed as they provided much needed traction.
Here, we had to alter the light shield. Throughout the last week of testing, it was removed and replaced multiple times, hence the larger amount of tape than last week's design. It was also trimmed down to hover just over the floor as it was dragging and prevented proper turning if there was any bump in the track.
We made some changes to the overall PID values in the code. We discovered that our line readings were so far from the wheels that any small turn from the back resulted in large sweeping movement of the photoresistors, especially when it came to oscillations when coming out of a turn. We doubled the derivative (Kd) coefficient and this was sufficient to smooth out some oscillations. We also played around with lowering our proportional and integral terms (kp and ki respectively), however this resulted in Tart not being able to make some turns. Doubling Kd ended up showing us the best results.
Interactive Robot Photo
11/Mar/2022
Hover over the yellow boxes to view our new insights. Most of Tart remained the same as last week, so for a more indepth glimpse into our design process, please refer to last week's interactive photo.
