Ichabod Crane: This cylindrical coordinate robot arm is designed much like a crane. But is is hump ugly, thus Ichabod seemed a good name to put in front of crane. It is constructed from aluminum bar stock available from Ace or Home Depot, and whatever motor or gearbox I could scrounge up. It's brains are provided by software control using an OOPIC microcontroller.

It was built to compete in the Find the Object and Remove It contest between the Denver Area Robotics Club and the Front Range Robotics Group based in Fort Collins, CO. Clik here to see rules of the contest and description. This robot is custom designed to do this contest and pretty much nothing else.

Essentially the idea for this design is a stationary robot arm that can rotate around the circular arena, "looking" for objects using an infrared ranging sensor (Sharp GP2D02 and GP2D12). When it finds one, it lowers the arm which is carried at highest position while in search mode to avoid accidental collisions. It then sends out a gripper mounted on a slide mechanism to locate the object at the range found by the ranging sensor. At this point, the robot won't know how high the object is, so it drops slowly until a downward looking sensor sees the object. This range is used to determine how high the object is. An object that is 4.5 or 6.5 inches high is to be left alone. All others are to be removed from the arena. So if the object is not one of those two "bad" objects, the arm drops, gripper grabs, and the object is slid out the arm to outside the arena and dropped.

I am most definitely not done yet. I haven't mounted the IR ranging sensors below the arm that will look out radially. I haven't convinced myself that the lift motor has enough juice to pick up the arm with an object. I think it will, but at the very least it will be slow which will hurt in terms of the 3 minute time limit.

It has 4 degrees of freedom viewable in the following subassemblies: 1) the rotating base, 2) the arm lift tower, 3) The slide, and 3) the gripper.

Rotating base: The rotation of the entire assembly is accomplished via the base unit which is split into 2 sections. The bottom section is where the rotating motor and the batteries are quartered. Batteries are removed for now during development. The stepper motor's output gear can be seen to penetrate into the next level up which is the gear box for the base. I made this gearbox by taking a gearbox from an old copier and then removing the gears and their shafts. I then took this plate which had holes where the shafts were mounted and used it as a template to make my own gearbox. The metal plates shown here were cut out of a computer case. I saw an old PC in a neighbor's trash some time ago and said "wow, look at that metal. I can use that". 6 months later or so, I did. The gearbox's output shaft penetrates into the next level where I use a set screw to made the shaft to the upper section of the lazy suzan onto which the arm assembly is bolted. In this photo you can just make out the bearings in the lazy suzan (Ace hardware, 4" square, $3.50).

Full view

Front view, on side with the arm

Lift tower: The main tower of the arm uses another stepper motor and gear box (taken from a scanner this time) to lift the entire arm assembly. The lift is done and kept stable and in one plane by lifting along 2 parallel brass columns that are lubricated with grease. The stepper motors force is transferred via a belt that is linked to the arm.

The actual arm extends 15 inches from the edge of the base (giving a 17+ inch "reach" to beyond the edge of the 30 inch arena). This long moment arm causes some hitching on the lift mechanism so I offset this a bit with a cable that helps in support. A counter weight (not shown in these photos as I'm still getting its weight) will sit on the stub of the back side of the arm to help counter the force of the arm. I can place batteries and any other ballast I need in the base. For once, weight is not really a consideration on a robot design. The OOPIC controller is visible in background. The arm is also broken into 2 pieces that can be diassassembled for easier transportation of the arm.

The Slide: The gripper has to be movable along the length of the arm. This is accomplished by the slide assembly. I tried using a scrounged stepper motor, but it was too heavy, so I used a modified servo. I did get to use a scrounged belt and sprocket, again from that yard sale scanner.

The downside to using the servo is that I don't have precise position contorl. I didn't have enough rotation with an unmodified servo, so I just plan to move it slowly and calibrate a minimum time to drive the motor which will equal a distance. I do this in software. Then I call that function with an argument that specifies how many of those distance "ticks" I want to move.

The gripper: I tried a few things on this before settling on a portable "noose" like configuration as shown here. A mini servo modified for continuous rotation turns a pulley that tightens or loosens the "noose" (some 28 guage stranded wire). It works surprisingly well, but boy is it ugly. The boxey frame is made with 1/16" aluminum to save on weight vice the 1/8" stuff everywhere else in the arm.

You can see the GP2D120 IR ranging sensor just above the yellow object and the belt of the slide mechanism.