Let's Make Robots!

Advanced Robot Arm Project

This is a long term project to develop a robotic arm that is more advanced that traditional hobbyist arms using RC servos without being so expensive that only well funded universities can afford it. 

The first prototype gripper used standard RC servos while we experimented with the configuration as shown in the first video. Once I had a good gripper configuration I built a prototype arm for it as shown in the second video. This arm used a differential drive system with the goal being to keep the weight of the motors as close to the shoulder as possible.

Although the differential drive did work, it's accuracy was poor. Even with the highest quality differential gears it was never going to have the precision I wanted.

Since then I have focussed on the gripper as this is the most difficult part with a lot of precision moving parts in a small space. The second prototype of the gripper still used RC servos but I tried using steel cables to link the servos to the fingers. Unfortunately the steel cables stretched more and snapped easier than expected.

You can see in this photo of the second prototype that pressure sensors were fitted to the finger tips and palm. These sensors used QTC as the sensing element. The finger joints use magnetic rotary encoders to measure angles. The third video shows you a simple test of the second prototype.

The second prototype was also the first to use molded rubber grips. The first prototype simply used a sample rubber sheet cut and stretched over the fingers for traction. The biggest problem here is that soft rubber is needed for good grip but when handling heavy objects the screws holding it in place can pull through the rubber. The rubber skin is not a major issue right now so it has been put on hold until the mechanical design is complete.

Recently I have been working on the third prototype. Unlike the previous prototypes this gripper uses 3-phase brushless motors and a minimal gear set. These motors are more efficient and have higher torque than standard DC brushed motors. I have kept gears to a minimum to reduce play between the gears. The gearboxes also have adjustments to minimize play further. All aluminium parts have been hollowed out to reduce weight as much as possible.

The third prototype uses an I2C bus for everything. Each motor controller and all magnetic rotary encoders are on the I2C bus to reduce the wire count and make the gripper easy to control using anything from a PC to an Arduino. In fact, each motor controller is a dedicated Arduino compatible controller to make the gripper easy to hack.

I have now started work on the 4th prototype. Unlike the previous versions this will be molded from polycarbonate rather than milled from aluminium. The polycarbonate is lighter than aluminium and it quite tough. It will bend and spring back into shape in situations where the aluminium would dent.

These images from Sketchup show the basic configurations of the gripper. This first configuration is ideal for pressing a button or flicking a switch.

This second configuration uses opposing fingers as a pair of tweezers for picking up small objects.

The third basic configuration is better for picking up cylindrical objects such as cups, cans and bottles. 

The fourth configuration can be used to pickup round objects such as a ball.

This is a work in progress so there will be updates.








Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.


this is really cool !

this 3 fingers hand is really versatile,  i also tought to go in the way of your 4th prototype.

maybe using smaller dc motor with a gearbox before the worm drive will give you more control and precision but less speed.


Precision is not a problem.

The brushless motors I'm experimenting with cog at 36 steps per revolution. As a result I can get the motor to step like a motor in 10° increments. The worm drive provides a 60:1 ratio so I end up with 1/6° resolution at the fingers. Because I only have these 2 gears and their distance can be adjusted for optimal meshing I have virtually no play.

Currently my biggest problem is that these motors are designed for model aircraft where they normally spin at very high speed. At low speeds they are essentially a dead short and draw very high currents in excess of 20A. Their windings have very low inductance so even high frequency PWM does not help much.

I need to rewind these motors so that I have more turns of a thinner guage wire. I also need to change them from a delta configuration to a star configuration.


Found this a couple of days ago, total opposite of your design but still, very handy :) for those who have access to a 3D printer.

Yale Open Source Underactuated Gripper 

That's not a bad design.

Totally forgot to ask above, what's the approximate weight of the whole thing?

I totally don't know. As the final gripper will be molded from polycarbonate rather than milled from aluminium I can't even guess.

I haven't seen many grippers with a design even remotely like this.  You've solved some big problems with elegance here.  I hope to see it in production and in the hands of universities soon!

I watched a demo a bit over a year ago done at an Unofficial Propeller Expo which used a similar looking robotic hand. Rick Galison uses the Propeller in some of his movie props and showed off one of Doc Ock's grippers. Apparently the gripper needed to be able to play an organ so all the fingers needed to be able to be configured in a line and the requirement for the gripper also to hold a pipe required it to be able to change configurations. Pretty darn cool IMO (like Oddbot's gripper).

I spent some time yesterday trying to find a video of Doc Ock's gripper without success. Rick does have a YouTube channel with some of his other gizmos displayed. He's made several pretty very cool paintball gattling guns.

Unfortunately it has ended up looking a lot like the Barrett robotic hand. This was not intentional, more a case of parallel evolution where form is dictated by function. I could not leave the motors lying flat on the palm like they were in the first version because it limited the range of motion too much.

Although I have made hands like this as a kid with legos it was this video here: http://www.youtube.com/watch?v=-KxjVlaLBmk that inspired the design.


Once again I'm blown away by one of your projects.

I have a CrustCrawler AX-12 SmartArm. I find it has a hard time picking up an empty soda can. Forget a full can.

When I first saw the three different configurations of the gripper, I assumed the positions would have to be manually set. I was very pleased to see the configuration change on the fly.

It looks like you're on to something great.

I hope there's a kit soon.