Week #10: Are We Engineers Yet?

Daniel, Jeremias - Gyro

Testing out Gyro to improve turning

we got it able to read and print stop when it hits 90.0 the overshoot is minimal, especially since we are turning it really quickly

we can also increase the checking frequency (currently its at .5 s)

All - Drive Test With Primitive Pick Up

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

It worked alright, but we need to re-calibrate a lot of things

Turning:

In grass, the turning is near perfect (as far as we can tell with our eyes)

One interesting thing we needed to do was change the turning distance measured by the gyro to count till 92 degrees have passed

It takes about 7 seconds to turn 90 degrees.

All - Implemented Stabby Array

This would give us depth for when we stab down to pick up the poop

Jeremias - Magnetometer + Gyroscope

Tried to combine the magnetometer and the gyroscope

Initially successful

Jeremias, Daniel - Mag + Gyro

There was a steady state error developed, roughly 20 degrees off

This is different from a few days ago when the mag was inconsistently off

We looked into the raw and processed values. We think that because there is a constant error, the detection is fine, we just maybe need to calibrate it or test it in multiple settings.

We think we need to try this outside where there is a potentially more constant magnetic field

Jeremias, Sailesh, Daniel - Mag + Gyro Test

Test outside and it is VERY unreliable. Sometimes its a perfect 90. Othertimes it turns for just a few degrees and stops. We are scrapping it.

Sailesh, Daniel, Jeremias - Test Run

Readjusted rear right motor to give for more space on the side

Inflating the rear wheels more to improve turning and speed

We tried this in different patches of grass and in the AstroTurf

our model

Sailesh, Daniel, Jeremias - IMA AstroTurf

Jeremias couldn't breath

Daniel got dizzy and had to go AFK

Sailesh was fine

The AstroTurf works really well!

• Driving was relatively straight. however, our movement is limited by our nut

• the turning is really good

• stabbing works! :)

Wi-Fi module is inconsistent. this causes the communication between the robot and computer to be glitchy at times. We have a patch fix working.

Battery issue: the battery pack supplying the arduino got too low. we are using the 12 V battery pack to supply the MPU now. The problem is we've already gone through 24 batteries...

Boundary Detection: worked pretty well. We will further optimize for white blocks.

Computer vision: working pretty well. it can detect the poop fairly reliably after we adjusted the HSV settings for the new setting. However, shadows are now a big issue,

Logical Issues: with our code. the state transitions are not optimal. For example, its especially crappy around alignment and boundaries

In theory, we can do this! we just need a TON of refinement

Sailesh, Jeremias, Daniel - WiFi Optimization

We need to speed up the rate of communication between the computer and the Elegoo because sometimes our robot moves too quickly and overshoots and rolls by the poop. We tried different setup tools, string, packets, etc.

We test drove at the AstroTurf again to get values for the computer vision (in particular the boundaries)

Sailesh -Quad Core Processor

Added Pi for on-board image processing

this allows us to not worry about the wifi issues we've been seeing earlier We don't need the WiFi chip or the voltage regulator anymore

Daniel - Video

Made a bomb video! million hits! https://drive.google.com/file/d/0BzokegCSY6auQ1JVRW5kLVBkSzA/view?usp=sharing

Daniel - Triangle

Finding the appropriate amount of error we can have

Sailesh, Daniel, Jeremias - Testing

We needed to retouch the RPi logic

• the camera would send all gray inital images which would cause the boundary detection to go off, so we needed to fix that

There is also an almost 5x5 space already demarked

• the only problem is that one line is the yellow line

SPI error: we needed to come back in to fix the camera

Sailesh, Daniel, Jeremias - Testing

fixed camera

camera detection is very reliable

worked on turning, logic is good

Gyro low battery

we work out

it's hot :(

also we are Poomba

Sailesh, Daniel, Jeremias - WERK

Redoo the wheels, with the metal nuts

test in darker lighting

Added more blades

Tested robot

Gyroscope wasn't working but everything else would theoretically work

the gyro the bike group gave us was slightly different so we

we also broke the elegoo's 3.3 V pin

Sailesh, Daniel, Jeremias - Please Work

got new gyro that looks the same but is apparently different enough that the code doesnt work too easily

after 3 different gyros, this one works mostly, has stead state error

It works half the time, the movement is severely limiting

We fixed everything else

we made the camera taller so that we have a wider field of the field

Gyro changes:

• increased increment from 10 to 50 then to 100

• developed a filter

We made made night filtered parameters as we were testing it in the late of night

• rain proofed our robot because weather is bad in June

• We spent $1600 + 50 a month on headlights

• yeah we tied our phones to the front

WEDNESDAY MIDNIGHT REPORT - IT WORKS

see you at 1030

Misc

Tighten motor bracket because its slipping

secure the stabby plank

Readjusted rear right motor to give for more space.

The front 3D printed nut broke so we had to go replace them

To Doo:

• Kickstarter

• explain what the user needs to use the poopEE

• comparisons

• system specs needed

• describe purpose