Autonomous Recycling Robot

[madsci1016]

Figured as I am becoming more involved in the SoR forums, i would post the only video i have of one of my robots. This was for my Undergraduate Senior Design project

[youtube]QpEDXo-Rw_c[/youtube]

More can be read about it here

http://www.billporter.info/projects/robot1.html

I intend on having a more proper writeup eventually, that whole website was created by me in 3 days because of a time crunch.

[galannthegreat]

Sweet 'bot man. I can see this being very useful for "party animals".

[SeagullOne]

Awesome! I can see these robots rolling around in city parks and whatnot gathering up recyclables at night after all the people have left.

The autonomous operation of this robot looks extremely well crafted. It seems to work so smartly given its 8' 8' foot environment.
The design looks catchy too!

I want to build one! 
You've got me thinking about what I could do with all the leftover parts that didn't work for my project, NINA.

Great job!

[madsci1016]

Thanks for all the kind words, lol.

My Role was Lead engineer, which meant i got to make all the technical decisions.
I also designed and built all the electrical aspects of the robot. I wasn't the lead programmer, but i ended up designing and programing the main search and acquire algorithm. (I whipped up a simple program just to test a new IR sensor and it ended up working way better then the method the programmer had been working on for days.) 

I also wrote the code that drives it around, which is tied into the search and acquire loop.

It's hard to believe that the only sensor it has to locate objects is one Sharp IR.

We actually tried the Admin's edge tracking code
http://www.societyofrobots.com/robot_sumo.shtml 
http://www.instructables.com/id/SGI9TTJF4WY1KW6/

 first, but it didn't work nearly as well as my code did (sorry Admin, no offense :-).

[Joker94]

great job, i can see a big future for that type of robot. And not in the to far distant future either.

[madsci1016]

Thanks!

[SmAsH]

That is so impressive!
I would buy one of those if they were on sale! Not so much as to use it... But the coolness factor too!
Great job man, really.

[madsci1016]

That is so impressive!
I would buy one of those if they were on sale! Not so much as to use it... But the coolness factor too!
Great job man, really.

The totally build bill came to around $1300 with some parts left over. Wasn't too cheap.

[Admin]

impressive!

first, but it didn't work nearly as well as my code did (sorry Admin, no offense :-).

lol yea probably right. I think the stampy algorithm works better for tracking moving objects.

[madsci1016]

For what we needed, it was slow and inaccurate at moderate traveling speed. (If i remember right) it was also driving the IR servo using delays instead of using the hardware PWM on the command module which caused shacking when we started adding our own interrupts.

FYI, the drive system and micro-controller is actually a chopped up Irobot Create and Command Module.

[Admin]

it was also driving the IR servo using delays instead of using the hardware PWM on the command module which caused shacking when we started adding our own interrupts.

Yea thats why I never use interrupts - it interferes with the servo PWM. The solution would be to use interrupts for the servo PWM as well, or just not use interrupts at all.

I found that tweaking thresholds was important to get the Stampy algorithm to work, too.

[madsci1016]

Your stampy code worked well, at slow speeds just like in your videos. We had a time restrictions, so we had to crank up the travel speeds (300 mm/s and faster), and at that speed, the robot would start behaving erratically, miss objects, turn the wrong way etc. We tried tweaking scan resolutions, timer and everything we could.

We ended up using one of the hardware PWM channels on the command module (atmega 168) and the IR servo worked fine after that, and we upgraded the Sharp IR to the bigger model that can scan father out. The code i ended up writing for the scanning algorithm worked better at high speeds by breaking up the field in front of it into zones, and as an object moved into different zone relative to the robot, the driving program would change (different speeds and turning delays). It would also ignore all contacts except for the closest one.

Your tut on stampy was great and taught us a lot, but it just wasn't able to do what we wanted.