Week #9: PoopEE

TL;DR:

We re-branded from The Pooper Scooper to PoopEE. The reason is because we are removing our scoop for a potentially more robust and effective method of picking up the poop.

We have been developing various methods to improve our turning

• rebuilding chassis to have a long horizontal axis

• integrating a magnetometer

Our mechanics for moving straight and turning are really good. We just need to calibrate it accordingly.

Daniel, Sailesh, Jeremias - Rebuild Chassis

Instead of having a vertical kind of built, with the length greater than the width, we made the width greater than the length. The advantages of having the wheels closer together on a side while extending the axles are that when turning, the wheels rotating in the same direction (aka the ones on the same side) are closer together and turn on similar paths. If those wheels were farther apart, the front wheels (the ones not powered by the motors), would be moved perpendicular.

http://www.rakeshmondal.info/4-Wheel-Drive-Robot-Design

Jeremias further cut the front axle.

Daniel and Sailesh bought 4 new brackets (to keep the front axle aligned)

We then assembled the new frame. We also had to get clever to fit the motors on the frame without having the wheels touch. (Bent the motor brackets).

Sailesh, Daniel, Jeremias, Han (All) - Tested Rebuilt Chassis

The turning works so well now! Because we were especially meticulous with alignment, as it has been a very compounding source of error, it is able to drive straight assuming the terrain is flat.

Daniel, Han - Scoop Test Smooth

The smoother print worked much better. The main difference is that the edge, instead of being a sharper kind of scoop which would get caught, is more circular. The corners are curved as well. We find that our scooper is much better in smooth, short grass because of this.

Sailesh, Daniel, Jeremias, Han (All) - Scoop Rake

We think that if we make the scoop more like a rake

It works okay-ish. Still gets caught in grass, but that is based upon angling. When its flat, there isn't much of an issue.

Sailesh - Computer Vision

Changed poop parameter

Added ellipses for poop

Daniel - Report

Wrote some of the report, threw in the general guidelines.

Developed on ideas and wrote general framework

Jeremias, Daniel - Magnetometer

Figured out how to use the compass

This would be instrumental for turning 90 degrees, which could be better than the timer

Jeremias, Sailesh, Daniel - Magnetometer

Though we were able to read the magnetometer and have proper logic, the sensor itself is just giving poor readings.

The 3-D printed nuts broke as well.

We are going to look into gyroscopes.

logic for compass based turning

• while ((curr > (init - turn)%360) || curr < init + turn)

• while ((curr > (init - turn)) || (curr < (init + turn)%360))

• while (curr > init - turn && curr < init + turn)

Sailesh, Daniel, Jeremias - Magnetometer

IT WORKS

We redid the calibration procedure and got our magnetometer to work. The x-offset and y-offset are now -33 and -170 respectively.

Our testing procedure is having it continually turn until it reaches 90 degrees. It has a much smaller error size now than it did yesterday. However, the error still persists. We think that if we integrate some self correcting algorithms, that would be able to mitigate compounding error.

• realigning every time it reaches the boundary

• recognizing the initial direction (initial being when it gets started up)

Sailesh - Computer Vision Refinement

improved color group detection for certain objects (see green circle)

Sailesh - Integrated Movement and Boundary Detection

When the robot approaches the boundary, it now makes a 180

Sailesh, Jeremias

Rear wheel motor brackets are very secure with metal nuts

All - Changing Fundamental Design

We are changing the overall design of the scoop

instead of picking up the poop from below, we are going to try to stab it from the side

Disadvantages of the scoop:

• relies on 3D printing

• Gets caught in the grass

• doesnt pick up the poop reliably

Advantages of the stab relative to the scoop:

• made out of wood and nails (sturdier, quicker and easier to make)

• doesn't get caught in grass

• Initial prototype was promising in effectiveness

• Kick the FireX roomba's robotic butt

Other Updates

Our 3-D printed nuts keep breaking. A lot of the newly printed ones don't fit.

Our scoop is some-what working-ish not really

Chassis can drive straight really well, turning is the main movement challenge remaining

We attached our motor to the front of the chassis

To Doo:

• Tighten motor bracket because its slipping

• Test the forward, detect, slow motor, down, forward, motor up (sailesh knows what I'm talking about)

• Secure stabby plank

• test gyro

• Kickstarter

• explain what the user needs to use the pooper scooper

• comparisons

• system specs needed

• describe purpose

• Decals

• LED

• Sirens

• Chrome Spray paint