We developed and implemented new tiny robots for swarm robotics research!

[beeboopbeep]

Hi everyone,

I just wanted to share the tiny robots that we designed and built to be used for swarm robotics research:

Few details about the robots:
- nRF51422 is the main SoC of the robot and ATmega328P is responsible for handling all the modulated IR related tasks
- robots communicate using mesh network via ANT Wireless
- 3.7 V Li-Po battery with a total of 120 mAh of capacity allows the robot to last up to 6 hours of operation
- individual robot estimates the proximities of nearby robots (up to 150 mm sensing radius) with the use of modulated IR signals and inputs the proximity data into the swarm algorithms for global behaviors to emerge
- this means that the robots operate in a fully decentralized manner (i.e., there is no one central brain that individually controls the robots)
- two small DC motors are used for actuation and the shafts of the motors are in direct contact with the floor and act as wheels (as you can imagine, a lot of slippages occur but the collective behavior algorithms are robust enough to deal with this)
- thanks to it's modular design, various sensors can be mounted onto the robots with direct access to the SoC's TWI/I2C, UART, and SPI

Thanks!

[ez24]

Interesting and fun to watch.  Hope you will post more.  Thanks

[mikeselectricstuff]

Many years ago I played with that sort of drive system - been meaning to do more but never quite got round to it - a small piece of silicone sleeving improves grip significantly

[beeboopbeep]

Hi Mike,

We tried several different wire heat shrink tubes but the motor shaft diameters were too small unfortunately (here is a link to the motors: https://www.precisionmicrodrives.com/product/104-001-4mm-dc-motor-8mm-type). What would happen with the heat shrink tube was that the cutoff ends, where the motor shaft ends, became too 'bumpy' because the tube shrinkage was not snug enough. This prevented the robot from moving forward due to lack of power. The other option we went with is putting a small dab of glue on the ends of the shafts with a heated glue gun and removed the excess glue with a soldering iron.

However, from the pictures you provided, the tube looks very snug on the motor shafts. May I ask what the diameter of the motor shafts were and what kind of tubing you used (if it isn't heat shrink tube)?

Thank you.

[timb]

Hi Mike,

We tried several different wire heat shrink tubes but the motor shaft diameters were too small unfortunately (here is a link to the motors: https://www.precisionmicrodrives.com/product/104-001-4mm-dc-motor-8mm-type). What would happen with the heat shrink tube was that the cutoff ends, where the motor shaft ends, became too 'bumpy' because the tube shrinkage was not snug enough. This prevented the robot from moving forward due to lack of power. The other option we went with is putting a small dab of glue on the ends of the shafts with a heated glue gun and removed the excess glue with a soldering iron.

However, from the pictures you provided, the tube looks very snug on the motor shafts. May I ask what the diameter of the motor shafts were and what kind of tubing you used (if it isn't heat shrink tube)?

Thank you.

You could also try dipping the shafts in that "liquid rubber" stuff you can buy at the hardware store. One brand is called Plasti Dip; it comes in paint cans and aerosol spray cans. A lot of people dip the handles of their tools (screw drivers, hammers, pliers, etc.) in it, to give them a nice rubberized grip.

[beeboopbeep]

Thank you Tim, I'll definitely have a look and experiment with it

[DaJMasta]

I was actually thinking the same thing, you can even do a couple of dips for a bit more thickness.  May be difficult to get exactly even between motors,  but should definitely wear better than just heat shrink.  Maybe making little rubber roller wheels would be another step better - I know there are certain mechanical pencil erasers that are only like 3-4mm wide, and even if that eraser is like 15mm long, that's still at least 5 wheels if you have a reliable way to poke a hole down the middle to get them on.

I've always liked the idea of swarm bots, programming behaviors for individuals or individuals on a network and then just seeing what comes out of it.  I know there's a similar thing I saw a year or two ago from a major tech university of some sort, but instead of roller wheels like this, they used a pair of vibration motors, which can give you more or less two axis movement running each at different speeds.

Looks like you have some decent horsepower to play with on each, how's developing for those chips?  Seems like a winner for this sort of networked low power application with the wireless built in.

[beeboopbeep]

I've always liked the idea of swarm bots, programming behaviors for individuals or individuals on a network and then just seeing what comes out of it.  I know there's a similar thing I saw a year or two ago from a major tech university of some sort, but instead of roller wheels like this, they used a pair of vibration motors, which can give you more or less two axis movement running each at different speeds.

... how's developing for those chips?  Seems like a winner for this sort of networked low power application with the wireless built in.

You're thinking of Kilobots from Harvard University with regards to the robots with vibration motors. Neat little robots and the designers have certainly met their design objectives of trying to be cheap and making it scalable (1024 units operated together on one workspace breaking world record). Here are the same vibration motors if you're interested in playing with them: https://www.sparkfun.com/products/8449

The nRF51422s were a lot of fun to program with but the learning curve was steeper than I expected, especially since I'm coming from programming ATtiny and ATmega chips where free resources from communities are unlimited. It certainly has all the wireless capabilities I needed for the project since it has Bluetooth Smart and ANT Wireless where they can both work simultaneously and allow point-to-point and mesh network configurations. Programming wasn't too bad either assuming you read all the necessary technical notes provided by Nordic on the nRF51 series which took few weeks. In addition, code can be compiled with ARM GCC so I was able to develop on Eclipse. Also, they provided all the necessary schematics and even layout guidelines for the circuit boards for wireless transmission (although there were plenty of third party vendors for the antenna filter hardware who provided the schematic and guidelines as well).

[mikeselectricstuff]

Hi Mike,

We tried several different wire heat shrink tubes but the motor shaft diameters were too small unfortunately (here is a link to the motors: https://www.precisionmicrodrives.com/product/104-001-4mm-dc-motor-8mm-type). What would happen with the heat shrink tube was that the cutoff ends, where the motor shaft ends, became too 'bumpy' because the tube shrinkage was not snug enough. This prevented the robot from moving forward due to lack of power. The other option we went with is putting a small dab of glue on the ends of the shafts with a heated glue gun and removed the excess glue with a soldering iron.

However, from the pictures you provided, the tube looks very snug on the motor shafts. May I ask what the diameter of the motor shafts were and what kind of tubing you used (if it isn't heat shrink tube)?

Thank you.

it wasn't heatshrink - it was silicone rubber, so held on by its own tension. Don't know exact dims - it was just some stuff I had lying around. Motors were normal pager motors so probably 1mm or so shaft dia.
If I was doing it again I'd look at moulding some little closed-end condom-shaped pieces from liquid silicone.