Trying to design a team-mini sumo robot and stuck! (not so stuck anymore, thanks to your ideas :) )

Community General Discussions
1

Hello guys,

I had been working on a project similar to a mini sumo robot, a "Team" Mini Sumo robot. I have a good PCB design for single mini sumo, which has Sharp 340K sensors/ Custom built IR sensors with TSSP4038 etc. It performed well in contests, but I haven't got a rank yet. Main reason is that in Turkey, where I live, there are lots of mini sumo builders because it's cheaper than a "Normal" Sumo and easier than micromice vice versa. But, there is a special branch of mini sumo, which needs some extra work on program and circuitry, is "Team mini sumo." It is a variation which instead of matching 1 vs 1, you match 3 vs 3. The robots in same team must recognize each other, so that they do not push themselves out. Teamkilling is punished with minus points generally.

So, I need a way to distinguish ally robots and enemy robots. I first thought of using IR signal, but this had some problems. First, my proximity sensors are also IR sensors, which works with 38kHZ IR pulses. I thought that if I modulate a message, instead of raw blinks, I might be able to recognize allies, but this is too much work for a mini sumo robot's brain I think. Especially I need at least 5 opponent sensors and 4 ally recognition sensors.

Then I thought that if I use a higher frequency transmitter and receiver, like 56kHZ IR receivers, and a second small chip like 12F675 which will only check the allies. But this is impractical, because another chip means that connection will be needed between two ICs. This will hurt the agility of my robots.

So what I need is a way to communicate 3 robots, informing them about the direction of allies, and not hurting or blinding my 38KHz IR proximity sensors. I am waiting for nice ideas.

Well, there is a product available from pololu, it is called Beacon Transceiver Pair. But it only supports one sender and one receiver. And the price is fairly high. I also need four pins to read from it, but I am almost out of pins.

the product: https://www.pololu.com/product/702

I added some sample videos and the rules section of the contest at METU, one of the most famous Turkish universities

sample contest and rules: http://odturobotgunleri.org.tr/en/pages/team-mini-sumo

http://odturobotgunleri.org.tr/en/documents/download/13 (pdf file)

Important update: 6 days to freedom. After my last exam, I am free to design and manufacture my dream bots! Meanwhile, you can inform me about your nice ideas. 

Update 15-07-2015

Good news, everyone! The exams that I have attended were very successful! This will be a huge morale boost for me.

I have decided that using color sensors and leds would be the best choice. I have designed a small board which will carry ams TCS3200 color sensor chip and the configuration resistors. I will attach this module to one of the CCP pins of PIC18F46k22. There will be two color sensors on the front side of mini sumo bot. I have chosen blue and green (just like home and away outfits of soccer players) because they probably would not effect the correct operation of IR sensors. I will simply change the color sensor if the opponent is using same method and same color as me. A jumper will toggle the onboard LEDs color.

And about the main board, cosmetics and less than %10 of routing is still waiting me.

In about 30 days, I will order the sensors and motor drivers from Digi-Key.

https://www.youtube.com/watch?v=kAaKt5Jz9Dk
2

WiFi might do you good?

Dear: Murdertrk

Maybe you could replace your proximity sensors with pings. If not you could replace your IR communications with wifi that might do you good. Either way these are good ideas so good luck!!!

From: Noah

3

I’d look at the inexpensive
I’d look at the inexpensive radio modules that you can get nowadays for communications. Be ware of import taxes from China though.

As for telling the difference between the teams, you could do something we humans do: put them in uniforms. Not literatly, but you could paint all of a team one color and use color sensors to tell who is on you side. Or you could use a colored LED with a set frequency. If you put the LED on the top, it should be visible from all around the robot. By modulating the signal you could send information back and forth.

If you want your team to know your orientation as well as position, try putting one color on the right and one color on the left. By figuring out which is which, you can tell a lot.

I don’t know team mini sumo rules, but Raspberry Pi 2 and the pi cam might be the only sensor you’d need for that. OpenCV is a wonderful and free bundle of software for dealing with computer vision.

Just thinking out loud.

4

Use HC-SR04 as distance

Use HC-SR04 as distance sensor, and put a bright led on the front. Wahahaha!>:D Colors are a good solution, even flexible. If other don’t use bright leds you can get away with some leds and photo resistances.Or even rfid tags, if it reads a friend tag stops pushing.

It would be interesting some kind of teamworking, to collaborate to push three together one opponent at time.

5

Ping sensors are huge and

Ping sensors are huge and harder to incorporate than IR sensors. They are also slower and need more time to interpret the signal. I can’t change the main board, and the sensors. The robot is about 25mm in height now and if I can, I would even reduce it. The ping sensors would hurt the stealth of my bots.

Painting robots would be risky because if the opponent has the same color, then both teams would struggle.

And the word ‘communication’ was not clear. I don’t want to use a wifi module or something, if its possible. I want the soultion as simple as it can be. I only want to determine that if my proximity sensors detect an object in front of the robot/at right/at left, I want to know whether that object is ally or not. Cooperation between two or more robots is a later problem, but it would be nice to do.

But, color led idea is not bad. I can even use bicolor leds and if opponent uses same technique, I can change the color with a switch. Just like home and away outfits in football/soccer! Where can I find those colored LEDs and the color detectors for them?

RFID would be a good idea as well. Can you explain more? What are the limitations and requirements?

PS: The rules are included on my first post. The link at the bottom is rules.

PS: I have a very important exam next month, unfortunately I will not have much time to read and answer posts here. But in 22th of july, I will even provide the specs of my robot. 

6

RFID

I’ve played a lot with RFID lately and biggest limitation is range: 2-5cm max. Tags and reader must be nearly parallel.

Reading is stopped or influenced by sorrounding metal, plastic is indifferent.The most common reader is 4x3cm.

Really fast reading of the first page, <10ms, but it can take as much as 2s to read all the content. The first section has an ID and some writable bytes.

7

Only if it was as high as 15

Only if it was as high as 15 cm… so color leds is the only option for now.

8

** LEDs**
I don’t know where you get electronics. I find eBay a good source for such things, but I don’t know the import tax situation. You may also have online stores in your country.

If you have an LED color match problem, you can use either a bi-color as you suggest or an RGB LED, but those take three driving pins and to change the color you’d need some way of entering the color manually.

Are you making all three robots or is it a team of three players each of which are making their own robot.

One thing that worked for someone here in sumo is to put a sound absorbing material around the critical areas of the robot so it was difficult to detect by sonar. It might work for a couple of matches until everybody else did it then you’d have to use a new trick. :slight_smile:

If you only need friend/foe, then all you need are the LEDs. I’d use one at the top that can be seen, but I’d also try to put four others in parallel at front, back, left, and right… The Nordic modules are too long-ranged for friend/foe id, unless somebody can figure out something I’ve missed. Maybe if it were tuned as low as possible output, then it might only broadcase a short difference.

I’ve never tried to do these things and am only starting to think seriously about swarm robots which have some of the same problems.

9

I can buy electronic

I will make all of them. Actually I have some ready bots now but they have normal mini sumo code on them. They have female headers for bluetooth and I can attach the beacon circuitry there. But I am not happy with those pcbs. I etched them myself and there are some flaws with both production and design. I am going to have better pcbs with a superior microcontroller and more features made in China, I will probably use SeeedStudio or Elecrow as fab-house.

I can buy electronic components inside and outside of Turkey. There are companies who handle the customs and other bureaucratic stuff for you when you order stuff from digikey and farnell. There are also local shops in my city who both sell hobbyist grade electronics and professional stuff.

I can drive the bicolor leds with logic ICs and transistors. It seems that I am going to vertically attach a 2x10cm scrap fr4, and place the board with receivers and leds on top of that platform. Normally, I would place them on the main board, but it might interfere with the IR sensors. Raising the centre of mass might be a problem, but we will see if it is going to be a huge one.

Well, nobody uses ultrasonic sensors for mini sumo robots in Turkey due to their latency and (relatively) complexity to use. Sharp sensors are mostly used and some other robots has industrial grade omron e32-d62 and other more expensive sensors. I rarely encounter homemade IR sensors.

10

IR is hard to defeat. I
IR is hard to defeat. I don’t know of this will work, but if you could lower the temp of the outside of your bot, then maybe it would look farther away to an ir sensor. You could either use really cold water in a flexible bag, or the flexible blue ice things they use for wounds.

Just thinking, ya know.

11

Using rgb LEDs wouldn’t require any pins.

You could have simple bank’s of resistors to control the color of each LED. The only requirement would be reasonably accurate resistors to manage the current. I don’t know how you will sense the colors is the only issue.

12

There are active RFID which

There are active RFID which can have up to one meter, but i couldn’t experiment with them.

13

What about a really bright

What about a really bright IR led to blind the sensor? Or pulsing at the same frequency to disrupt measurements. Seeing the rules there’s nothing against it( the referee may still ban them anyway).

14

Duane is right, the sensors

Duane is right, the sensors work with the principle of sending a pulse and checking whether it returns or not. 

15

I will first test with a

I will first test with a dial potentiometer to ensure range, then I will place a fixed resistor. That’s what I do with my homemade IR sensors. And I will probably use some color sensors from TAOS or another company. They have SO-6/SO-8 packages which is tiny enough for me. I will draw some breakout boards and etch them myself. They will be one sided.

16

Guys, I need 22 days to

Guys, I need 22 days to start this huge project!

Meanwhile, I would appreciate any nice, not expensive dc motor ideas. I am currently using 2x or 4x pololu hp micromotors at 400 rpm or 625 rpm. (Depending on wheel diameter and other things) but they are not as powerful as faulhaber or maxon motors. Unfortunately, those motors from original distributors are fairly expensive. For each motor, they want as much as 250 usd. And the USD is expensive in my home currency (1$ = 2.7TL). 

My requirements are:

Size: max. Robot width allowed is 10 cm, and in order to comply with that, motors plus wheels should be less than 4.7cm (we need some place for wiring as well). With gearhead, I prefer less than 35mm, but if the diameter is less than 15mm, I can hide the motor into wheels. I machine wheels with cnc lathe.

Speed and torque: The optimal speed can be as low as 400 rpm and as high as 1000 rpm. Torque shouldn’t be lower than pololu hp 625rpm (sorry have no time to add specifications for now,  I might add them later).

Voltage and current: my current motor driver allows to draw 3.5A between 10 and 20 volts. If i am going to use 4 motors total, I will connect them in parallel. I supply the driver with 500 mah 3s 25c li-poly batteries without any regulation. (If you don’t count smoothing caps). That makes 12.5V when the battery is almost full. I might overvoltage the motors a bit, I already do it.

Motors with price about 20-50$ (gearhead inclusive) are appreciated. If your recommendation is more expensive, feel free to write but warn me that it’s going to hurt)

 

17

Have you considered a 4 wheel base

with one motor driving the front left and the other driving the back right? It would open your options for motors as long as you have the space to give up.

18

4 wheel is a problem when

4 wheel is a problem when you try to adjust the front ‘blade’. If those wheels are not connected to a motor, it would pe pointless to have them. I might place the motors vertically and use a crown gear to both reduce the speed and to change spinning orientation. That way, would I lose much power? 

19

finally free!

Hey guys,

 

I have successfully given my exams and I am ready to fiddle with my bots! I need a day or two to repair my damaged/improperly constructed bots. They might have this team mini sumo functionality too. I havent decided yet.

After that I will post current status of my schematics and robot bodies. (They will not be the design files because I still have to do lots of things.) Boards have tqfp components and other stuff, so they will be produced within a fab house.