Strategies for Sumobot design

jmokie

Hi, I am building my first sumobot, starting with a BoeBot controller.· I have purchased two new gear motors with 3" Wheels for the drive.· I will use a H-Bridge to interface the motors to the Controller.
What I would like to know is what is the minimum sensor configuration that would produce a good competitor.· Here's what I have planned:
4 QRB 1134 IR sensors, 2 front, 2 rear
2 IR Transceivers on front for object detection
Questions:
1.· Enough line sensors?· I have seen some with 3-4 sensors on each end.
2:· IR Transceivers, I have used the BoeBot kit separate emitters and detectors and see that i can detect and track an object with those, but need something more mechanically stable.....Suggestions?· Is two on the front enough?
Jerry
p.s. I am posting this on both this forum and the old yahoo forum to see which brings the most/best answers fastest.

macman

Hi,
I read your posting and I must say I am amazed that it did not get many replies, considering that Parallax make a Sumobot as well as the BoeBot. Unfortunatley I am a newbie myself so for me to give contruction or code advice would only make matters worse, but for what it is worth, you could just take a look at the Parallax code for the Sumobot that could give you a few ideas or another possibility is the Sumovore by Solarbotics, http://www.solarbotics.com/ -- The Sumovore kit runs sans microcontroller but can be adapted to run with a BS2 by using the Sumovore BS2 Brainboard Add-on. as a result they also have code posted for the BS2 which again may well be worth looking at.

The Sumovore has done very well for itsellf, indeed for an off the shelf kit it is very competative, just look at any Sumo contest results to verify. Even some of the competative customized Sumobots are adapted Sumovores, with upgrades to motors and wheels/tires. So you could use their hardware configuartion as a benchmark for your bot ; It uses two DC motors powered by a L293D motor driver with four opto-reflector edge sensors with an optional fifth edge sensor (when using a microprocessor brainboard) along with two 38kHz IR emitters and detectors for opponent detection. If you start with something similar I don't think you would be too far off base with your project.

Remember that traction is key when it comes to the pushing aspect of contact, so attention to tire composition can pay dividends, also don't forget tire surface area. One should also factor in that side contact can impede your power if your wheel is being pushed/lifted. Some Sumobot builders make the front panel reflective to draw their opponent into a frontal assault and hopefully discourage attacks to the sides or rear. The small wedge design seems to be popular right now, this design allows for a low center of gravity and is often combined with large tire surface area . But there is no definitive design as yet. The contest rules allow quite a lot of lattitude for inovative design so read the rules carefully maybe you might come up with a winning design/strategy.Please let us know how you make out with your Sumobot, I am also looking into a similar project also, so I for one am interested in your endeavors.


Good luck with the 'bot'
Regards Nigel

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards Nigel

Success is the ability to go from one failure to another with no loss of enthusiasm.
Winston Churchill

Jon Williams

We equipped the SumoBot with a nice compliment of sensors, but there are successful bots with less. At a minimum, one would think you'd like at least one IR detector to tell you the opponent is in front of you (we use two so that you can hunt), and that you'd like at least one line sensor so that you don't drive yourself out of the ring (again, the Parallax SumoBot uses two fo better control).

The nice thing about the mini Sumo class is that within the defined mechanical rules, you can do what you think is best. Do a web search and you will find an incredible variety of successful Sumo designs.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jon Williams
Applications Engineer, Parallax
Dallas Office

macman

If I remember correctly Jon did you not write sometime back that you are working on new code for the Parallax Sumbot?

If so what prompted the revision, may one ask?

Regards Nigel

jmokie

Nigel, Jon
Thanks for your replies. I had expected more of a debate on the best sensors, arrangement, etc., but I guess the experts don't want to give away secrets.

I have reviewed almost everything I can find on the web, but don't find much discussion about how to handle(with sensors), side attacks, rear attacks, etc. Since stamp doesn't have interrupts, the strategy of the code, i.e. switching from attack to defense, evasion, and such like would seem to have proponents for one approach or the other.

I'll post my set of strategies when I finish.
Thanks Guys

Jon Williams

macman said...
If I remember correctly Jon did you not write sometime back that you are working on new code for the Parallax Sumbot?

If so what prompted the revision, may one ask?

We updated the code for 2.5 syntax -- that's all.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jon Williams
Applications Engineer, Parallax
Dallas Office