Rack-and-Pinion [Type] Steering

Notes: An offshoot of this concept should make very accurate shaft- or gear-position sensing possible without using a rotation sensor. See the notes below.

Carl Jagt has modified and improved upon this concept. I’ve posted his pictures and text here.

It Works at Last!

I’ve been tinkering with this assembly for well over a month. The original concept was to get a robot to steer using a rack-and-pinion (R&P) steering arrangement—without the use of a single touch sensor. This concept was modified as outlined in the following notes.

Development: From Rubber-Band Pulley Drives to Pistons ...

The R&P system turned out to be extremely difficult to implement without using touch sensors or a position sensor. My initial idea was to use a rubber-band drive on a true rack-and-pinion steering system. The final design does not use either rack or pinion, but does steer in the same way.

The rubber-band concept used a pulley system to turn the pinion gear, and more rubber bands to recenter the wheels automatically when the steering motor was not actually running (the NQC float command allowed the motor to turn more easily when while the motor was off). The problem was that  rubber bands strong/tight enough to return the steering assembly to its center position were too strong to allow the belt-driven steering to move completely to the left or right, and vice-versa.

Another difficulty that surfaced was the strength of the steering assembly itself. This went through several revisions, with the final assembly using fairly standard Technic pieces as above. Note the gray, double-bored 2 x 1 bricks which the axles go through (they came with the Research Sub Technic set)—this allows the axles to be directly under the pivot point of the steering assembly.

After continually running into dead ends with a true rack-and-pinion steering system, I awoke one morning realizing that the same mechanical system that makes pistons work could be used to convert rotary motion to back-and-forth linear motion. This proved to be an excellent method of getting the steering to work. An added advantage is that the attachment point on the horizontal shaft (see photo) is easily adjusted.

The Touch Sensor as a Rotation (Position) Sensor

The next question to be answered was how the robot and its programs would be able to determine the position of the wheels. I had tried to develop a system that did not use any touch sensors, but fairly early on abandoned that as impractical, and focused on designing a system that uses only one touch sensor. By mounting a cam on the primary vertical shaft which strikes the touch sensor once each revolution (as pictured above), the program can count the rotations of the shaft, and, in conjunction with knowing the direction of the motor, easily derive the wheel position. There may be a further application of this—using a more significantly geared down system, the same touch sensor/cam arrangement ought to allow angle of rotation determination without the rotation sensor, with only minor inaccuracies due to gear lash.

Final Details

In the “finished” ’bot, a single drive wheel  in the rear provides the forward or reverse motion, which eliminates the need for a differential gear. Not included in the picture above is the steering motor, which has an 8t gear mounted to the large gear at the top of the photo. Sorry about not having any more photos—I still don’t own a digital camera, and am limited by what will scan on my flatbed scanner.

Lego, Lego MindStorms, and Technic, as well as some graphics used herein are all Copyright © TLG (The Lego Group).
All other contents Copyright © 1999, Douglas M. Wilcox—Reproduction of images and text is welcomed, so long as credit is given.