My Homemade Robot Arm: Part 1, the one that didn't make it

I have for some time been fascinated by the idea of building a robotic arm. I looked for a long time to find something already built and then to modify it, but even ebay specials always ran into a hundred or more dollars. You can buy one from OWI new for about the same, so finally I just said enough. I'll just build one myself from scratch.

As much as possible I used motors and chains and belts and other parts from stripped down equipment. Many of the parts for this beauty came from a 1960's copier I rescued from the trash. Although when I was done taking it apart that's where it ended up anyway.

The basic design concept is shown in this diagram. It has 5 degrees of freedom. A pivot motor allowing the whole thing to rotate. The shoulder joint acts much as a human shoulder, raising and lowering the arm. The elbow is easy enough to see. The wrist which both flexes up and down, and also rotates through about 200 degrees. And finally the gripper.

 

This photo shows each of the major joints with a number. A number with "a" next to it shows the motor that provides force for that joint from a distance. The elbow motor is a standard servo at 2a and uses a chain to translate force to a sprocket which then transfers that force again up to the elbow joint. The wrist flexion motor at 3a uses a belt and sprockets I took from a dead printer. It also is a standard servo. A micro servo is the force behind the gripper at 5.

DC motors provide force for the shoulder joint at 1, wrist rotation at 4. Not shown is a rotating base motor that will also be a DC type.

The biggest mistake I made on this baby was not picking a motor that had enough torque at the shoulder. In the first generation seen in the opening photo you can see I was using a belt/pulley system. The belt was hand made from some inner tube rubber. It stretched too much and didn't lift enough. So I redid the mounting bracket and bolted the arm straight to the big pulley then used a set screw to put that pulley directly on the motor. The motor had a gear box but it just wasn't strong enough. So I have to replace it.

In this photo you can also see the transfer of force chain/sprocket for the elbow. This allows me to control the elbow regardless of what angle the shoulder is at. I tried to keep the motors in the base and transfer force with chains, belts. I didn't do it enough though.

The elbow servo motor wasn't powerful enough either. I was able to alleviate that somewhat by using springs to help hold the elbow in a neutral position. This worked okay but made for a funny control system and I got some funky oscillations during movement. I don't recommend this.

The wrist and hand I was pretty pleased with. Fairly. The servo motor mounted midway down the forearm needs to get moved down the base somehow. I'm not sure exactly how to do this. With the really beefy shoulder motor I'm looking at using, it may not be necessary. But ideally to be more efficient that's how it should be done. I have removed the belt that provides motion for wrist flexion for clarity. You can see it in the main photo.

This photo shows the wrist at its max downward flexion with my own hand as a model. You can see the micro servo sticking out of the back of the hand that provides motion for the gripper.

This view shows the small dc gearhead motor that provides force for wrist rotation. The metal shaft next to it is the pivot point for wrist flexion. The large spur gear you see is not a gear but rather a bearing in which is mounted the shaft for the rotation shaft. I chose that spur gear because it had the bushing in it already and matched the shaft I wanted to use. Notice the two screws holding that spur gear in place.

The wrist rotation shaft I took out of a relatively new printer. The black gear is permanently fixed to the shaft and meshes with the small output gear of the DC motor in the wrist assembly. The photo below shows the wrist rotation motor meshing with the black gear.

You can see here that only one side of the gripper moves by way of the blue micro servo to which it is bolted. The other (right) "finger" is fixed. Crude but fairly effective. And quicker to make.

 

Here's another view of the wrist. You can see here is contained in a bracket of black painted metal with a sticker on it. The sticker has german and chinese characters on it so I couldn't tell you what it says.

Here you can see the black spur gear meshing with the small DC gearhead motor right under the sticker. I can rotate the wrist through about 340 degrees, more than a human hand. I could rotate more than that but would run the risk of fouling the wires for the gripper servo.

I intend to use rotational pots on the rotating joints for a feedback system to know their position. Limit switches on the gripper

 

click here to see a short video clip of each of the joints in motion. I only had them going one at a time at best.