Introduction: Build a Very Small Robot: Make the World's Smallest Wheeled Robot With a Gripper.

About: I believe that the purpose of life is to learn how to do our best and not give in to the weaker way.
Build a 1/20 cubic inch robot with a gripper that can pick up and move small objects. It is controlled by a Picaxe microcontroller. At this point in time, I believe this may be the world's smallest wheeled robot with a gripper. That will no doubt change, tomorrow or next week, when someone builds something smaller.

The main problem with building really small robots is the relatively large size of even the smallest motors and batteries. They take up most of the volume of a micro robot. I am experimenting with ways to eventually make robots that are truly microscopic. As an interim step, I made the three tiny robots and the controller described in this instructable. I believe with modifications, these proof of concept robots, could be scaled down to microscopic size.

After years of building small robots (see here: https://www.instructables.com/id/Building-Small-Robots-Making-One-Cubic-Inch-Micro/) , I decided the only way to make the smallest robots possible, was to have the motors, batteries, and even the Picaxe microcontroller external to the robot.

pic 1 shows R-20 a 1/20 cubic inch robot on a dime.

pic 1b and 1c shows smallest wheeled robot lifting and holding an 8 pin IC.

THERE IS A VIDEO in step 3 that shows the robot picking up an 8 pin IC and moving it.
And another video in step 5 that shows the robot turning on a dime.

Step 1: Tools and Materials

18x Picaxe microcontroller from Sparkfun: http://www.sparkfun.com/

Micro serial servo controller available from Polulu: http://www.pololu.com/

2 high torque servos from Polulu

2 standard servos from Polulu

.oo5" thick copper, brass, or phosphor bronze sheet metal from Micromark

2- 1/8" x 1/16" neodymium magnets

1- 1"x1"x1" neodymium magnet. Magnets available from: http://www.amazingmagnets.com/index.asp

Telescoping brass tubing from Micromark: http://www.micromark.com/

Brass pins from Walmart

Glass beads from Walmart

1/10" fiberglass circuit board material from Electronic Goldmine: http://www.goldmine-elec-products.com/

clear five minute epoxy

Assorted nuts and bolts

TOOLS

needle
tin snips
soldering iron
drill
metal files
small needle nose pliers

Pic 2 shows the Picaxe module used.

Pic 2b shows the back of the Picaxe module.

Step 2: Build a 1/20 Cubic Inch Robot.

At .40"x.50"x.46" the robot volume of the Magbot R-20 is slightly less than 1/20 of a cubic inch. It is made by folding 3 box structures of non magnetic sheet metal. The smallest inner box is soldered to the left finger of the gripper. two small magnets are epoxied to the vertical shaft which bends to form the right finger of the gripper which rotates freely. It is these two magnets which are controlled by an external moving rotating and spinning magnetic field that provide all the power to the robot.

I used .005" thick phosphor bronze sheet metal for the box structures because it can be soldered and dousnt oxidise or tarnish easily. Copper or brass could also be used. I originally used small drill bits to drill the bearing holes in the sheet metal for the rotating wire shafts. After breaking a few of them in a drill press, I ended up just punching holes with a large needle and hammer into the sheet metal. This creates a cone shape hole which can then be filed flat. The holes do not have to be a precise size or even perfectly placed. At this small a scale, the frictional forces are minute and if you look closely at the pictures you will see I used long .1" standard long header pins which are square, for the shafts and gripper fingers. Copper wire could also be used.

The glass bead wheels were mounted on brass pins epoxied to the bottom of the robot. It is important to use non magnetic materials for the construction or the power and control of the robot will be adversely affected.

Step 3: A Robot Magnetic Motor

The robot has four degrees of freedom. It can go forward and back, rotate left or right, move the gripper up and down, and open and close the gripper.

Pic 4- I relocated the four on board motors that it would normally take to do this by simply suspending a magnet horizontally on a two axis gimbal. Two 1/8"x1/8"x1/16" magnets are epoxied to a vertical shaft of wire that is bent to form one finger of the gripper. The two magnets are lined up to act as one magnet and create a single magnet motor. This is mounted in the smallest box which has the other gripper finger soldered to it.

The gripper box is mounted to the second horizontal axis of the gimbal with a 000 brass screw and nut. I used the screw so I could easily take it apart for adjustments.

An external magnetic field is mounted on a CNC type machine which can slide the magnetic feild along the x and y axis and rotate it horizontally and vertically. It could have been done with an electro magnet, but I chose to use a one cubic inch neodymium permanent magnet because it is the easiest and quickest way to create a large magnetic field in a small volume.

Pic 4c- So, with the north end of the tiny magnet in the robot facing toward the larger external south end of the magnet below it, the robot magnet follows fairly closely the motions of the external magnetic field.

For a short video of the robot picking up a 8 pin IC, see here: http://www.youtube.com/watch?v=uFh9SrXJ1EA

Or click on video below.

http://www.youtube.com/watch?v=uFh9SrXJ1EA

Step 4: CNC Type Robot Controller

Pic 5 shows the CNC type robot controller. Four servos provide motions to the one cubic inch neodymium magnet which the gimbal mounted magnet in the robot follows.

For the x and Y axis a high torque servo with a pulley and fishing leader pulls on the fiberglass platform. A spring opposes the motion. The platform rests on two telescoping brass tubes that act as a linear guide. Plastic bearings made from a plastic cutting board, on either side of the linear guides, keep the platform level.

This particular robot controller has a limited range of a few cubic inches. This should eventually prove more than adequate to control truly microscopic robots which may only require a range of a few cubic centimeters.

Step 5: Magnetic Robot Circuit

The robot controller consists of a Picaxe microcontroller which is programmed to provide a sequence of motions to the robot. I find the Picaxe to be the easiest and fastest microcontroller to hook up and program. While it is slower than a standard Pic Micro or Arduino, it is more than fast enough for most experimental robots.

For other Picaxe projects see here: http://www.inklesspress.com/picaxe_projects.htm

And here: https://www.instructables.com/id/Building-Small-Robots-Making-One-Cubic-Inch-Micro/

The Picaxe controls the robot by serially sending commands to a Polulu micro serial servo controller. The Polulu controller is very small and will continuously hold up to 8 servos in whatever position they are put in. Simple commands from the Picaxe allow you to easily control the position, speed and direction of the servos. I would highly recommend this controller for all kinds of servo based robots.

The schematic shows how the four servos are connected. Servo 0 and 1 guide the 1" magnet along the X and Y axis. Servo 2 is a continuous rotating servo that can rotate the magnet more than 360 degrees. Servo 3 tilts the magnet slightly forward and backward to lower and lift the gripper.

For a short video of the robot turning on a dime, see here: http://www.youtube.com/watch?v=wwT0wW-srYg

Or click video below:



Step 6: Robot Controller Software

Here is the software program for the Picaxe microcontroller. It sends pre-programmed sequences to the Polulu servo controller which moves the magnet in 3d space to control the robot. With slight modifications, it could also be used to program a Basic Stamp two.

To program the Picaxe I found it necessary to disconnect Pin 3 (serial output) from the servo controller. Otherwise the program would not download from the PC. I also found it necessary to disconnect pin three from the servo controller when turning the circuits on, to prevent the servo controller from locking up. Then, after a second or so I reconnected pin 3.


'Program for R-20 magrobot pickup sequence using a polulu servo controller

high 3 'serial output pin
pause 7000

'set to 0 position
serout 3,t2400,($80,$01,$04,1,35,127) 'position s1 13-24-35 counter-clockwise
serout 3,t2400,($80,$01,$04,0,35,127) 'position s0 c-clock
pause 7000

'level magnet
serout 3,t2400,($80,$01,$04,3,23,127) 'position mid
pause 1000

'move forward long servo1
serout 3,t2400,($80,$01,$04,1,21,127) 'position clockwise
pause 1500

'grip down
serout 3,t2400,($80,$01,$04,3,26,127) 'position down
pause 2000

'close grip
serout 3,t2400,($80,$01,$04,2,25,1) 'slow speed clock
pause 50

serout 3,t2400,($80,$01,$00,2,0,127) 'stop servo 2 rotate
pause 700

'move forward short
serout 3,t2400,($80,$01,$04,1,13,127) 'position clock
pause 1000

'grip up
serout 3,t2400,($80,$01,$04,3,23,127) 'position midpoint
pause 700

'turn right 90
serout 3,t2400,($80,$01,$04,2,25,1) 'slow speed clock
pause 470

serout 3,t2400,($80,$01,$00,2,0,127) 'stop servo 2 rotation
pause 1000

'forward
serout 3,t2400,($80,$01,$04,0,13,12) 'position s0
pause 1500

'grip down
serout 3,t2400,($80,$01,$04,3,25,12) 'position mid
pause 2000

'close grip
serout 3,t2400,($80,$01,$04,2,25,1) 'slow speed c-clockwise
pause 50
serout 3,t2400,($80,$01,$00,2,0,127) 'stop servo 2 rotation
pause 400

'backup
serout 3,t2400,($80,$01,$04,0,35,127) 'position s0 c-clock
pause 700

'grip up
serout 3,t2400,($80,$01,$04,3,22,12) 'position mid
pause 1000

pause 6000

'set to 0 position
serout 3,t2400,($80,$01,$04,1,35,127) 'position s1 13-24-35 c-clock
serout 3,t2400,($80,$01,$04,0,35,127) 'position s0 c-clock

loop:
goto loop

Step 7: Adding Sensors

This robot has no sensors. To be truly useful as a robot manipulator of small objects it would be an advantage to have a feedback loop to the microcontroller from various real world sensors. To avoid putting a power supply on board, light sensors could be used. Laser or infrared light could be directed to the top of the robot and mechanical reflectors or blockers could be connected to touch sensors, pressure sensors, or temperature sensors and variable reflectance read by photocells or a video camera.

Another possibility is to use RFID technology to transmit a pulse that powers electronics on the robot to return instead of an identifying number, a sequence of bits that represent variations in touch or other sensors.

Step 8: Other Magnetically Powered Robots

Robots controlled by magnetic fields of various types are nothing new. Some of them are microscopic and some are larger so they can be deployed medically in a human body. Some use computer controlled electromagnets and some use movable permanent magnets. Here are some links to some of the best and smallest experimental magnetic robots researchers are working on.

Flying magnet robot on a penny.
While it doesn't actually fly, it hovers in a computer controlled magnetic field, much like those toys that suspend a small globe of the earth. It also has a gripper that expands when heated with a laser and then grips as it cools. Unfortunately, the robots magnetic north and south ends are vertical, so there is no way to control the rotational spin to precisely orient the gripper. It is slightly larger than the smallest robot I made which is shown in step 9.

http://www.sciencedaily.com/releases/2009/04/090413205339.htm
http://news.cnet.com/8301-11386_3-10216870-76.html

Swimming magnet robot
A truly microscopic robot that is a spiral with a magnet at one end. With an external pivoting and rotating magnetic field, it can be aimed in any direction and swim underwater.
http://www.sciencedaily.com/releases/2009/04/090418085333.htm

Steerable camera pill by magnets.
http://www.spectrum.ieee.org/aug08/6469

Medical robots.
http://www.medindia.net/news/view_news_main.asp?x=5464

Magnetically controlled camera.
http://www.upi.com/Science_News/2008/06/05/Controlled_pill_camera_is_created/UPI-60051212691495/

Here are some microscopic magnetically controlled grippers that can be chemically or heat activated.
http://www.sciencedaily.com/releases/2009/01/090114210651.htm
Unfortunately, these micro grippers cannot release once they grab. So they are more like a microscopic bear trap than a fully functional gripper.
http://www.sciencedaily.com/releases/2009/01/090112201137.htm
http://www.rsc.org/chemistryworld/News/2009/January/13010901.asp

pic 10 shows the Magbots R-19, R-20 and R-21, the three robots I made for these experiments. The smallest one was made smaller by eliminating one pivot and the wheels. A wire tail keeps it from tipping over backwards.

Step 9: Building Even Smaller Robots

Pic 11 shows the Magbot R-21, the smallest magnetically powered robot with a functional gripper I have made so far. At .22"x.20"x.25" it is around 1/100 of a cubic inch. By eliminating the wheels and one pivot point (gimbal), the robot is much smaller than the wheeled version. It slides on the metal frame not quite as smoothly as the one with wheels. The wire tail allows the robot to rock back to lift the gripper.

Such a configuration could be used to create a microscopic sized robot. The problem at this point, is to either use conventional IC technology to create thin film mechanical structures, or to come up with some other alternative for creating microscopic structures. I am working on it.


These small robots represent one of the easiest ways to get a lot of motion in a small space. There are many other possible configurations of on board magnets and external magnetic fields that could produce very interesting robots. For example, using more than three or more rotating or pivoting magnets on a robot, could result in more degrees of freedom and more precise manipulation of the gripper.