The purpose of this tutorial is to walk through the specifications of the Micromouse maze, discuss the design of the maze, provide guidance in it construction, and furnish technical drawings for the maze.
Based on our records, the maze cost approximately $1820.00 to initially build in 2008/2009. Another $200 was spent to add legs and leveling feet to the maze in 2010.
The maze took about a quarter to build, working about 30-40 man hours on the weekends. We learned a lot during the build and wasted a lot of time due to improvised and on the fly planning. Doing the maze build again, I estimate it would take about a month of on the weekend building to complete the maze. I would not build my own walls, I would buy them. Active Robots Store is out of the UK, so the shipping will be pretty high. The main motivation for buying versus making the walls, is the time needed to make the walls is quite high and number of setups required to make walls is the highest of the build. Furthermore, making the walls requires more tools.
For comparison, the following link is for a complete off-the-shelf maze, with walls
Links to two companies sell maze walls and corner posts
CAMM Maze Specifications
IEEE Region 6 Maze Specifications
The CAMM and Region 6 specifications are basically the same, since CAMM is derived from Region 6's rules. The Maze that IEEE UCSD built in 2008/2009 is designed to comply with the Region 6 specifications. The differences between the CAMM and Region 6 specifications are in the handling of gaps, dents, discontinuities between the sheets of the maze sections.
The maze is designed in a modular fashion to allow the user to move the maze in a reasonable sized vehicle, store the maze in a reasonable sized space, and most importantly, move the maze through a normal sized door. Although there are doors that are over ten feet wide or tall, the door to the lab where our maze lives is not, so the ability to move the maze from where it was built to where it lives was a paramount motivation in the design of the maze.
I also spent a considerable amount of time placing the seams of the maze modules. The module seams are the primary source of the Mice getting hung up, since the seams are where virtually all of the discontinuities occur. I placed the seams as close to the edge of the cells as possible, so that the seam was not running down the center of the cell. I constrained the possible sizes of each cell to be less than the size of a standard sheet of plywood. The following table shows the measurements for the sheet of plywood used in each type of module.
|Measurement||Square Module||Small Rectangular Module||Large Rectangular Module||Standard Sheet Of Plywood (for reference)|
Other variants that use smaller modules are shown in the table below:
|Measurement||Rectangular Module||Square Module|
I wanted to use standard dimensional lumber to create the frame under the maze. I selected 2×4's because they are practically ubiquitous and I knew their nominal dimensions off hand. In hind sight, using normal dimensional lumber was a bad idea. I would use engineered 2×4's or create custom frame pieces from plywood. It proved to be too difficult to find normal 2×4's that were straight, square, and true, especially with 10' long 2×4's. If and/or when I build another maze, I will not be using standard dimensional lumber.
Holding the modules together tightly is a critical step in limiting surface discontinuities of the maze surface. Looking at the CAD model, the maze can be broken up into 3 major sub-assemblies.
There are 2 types of the major 3 sub-assemblies. There are two Edge Assemblies, one on the left and one on the right hand side of the maze shown in the rendering above. The Spine Assembly is seen between the two Edge Assemblies. See the table below for the composition of the sub-assemblies.
|Sub-assembly||Square Module (QTY)||Small Rectangular Module (QTY)||Large Rectangular Module (QTY)||QTY Used In Whole Maze|
Each sub-assembly is bolted together with shorter bolts, like the ones shown below.
More on these shorter bolts later.
Then each Edge Assemblies is bolted to the Spine Assembly with 6 bolts. Before the maze had legs, the bolts to hold the sub-assemblies together had to be several feet long. They were made with a long threaded rod, a hex head bolt, a coupling nut and thread locker. The photo below shows a pair of these bolts.
This photo shows the long bolts installed on the maze.
The thread locker is used to prevent the coupling nut from being unscrewed from either the threaded rod or the hex bolt. Since we built the maze, several of the thread locker joints have failed and required reapplication of the thread locking compound to the threads.
The shorter bolts shown in the picture above are carriage bolts. These bolts have a square shoulder under the head of the bolt and a rounded over head (See Wikipedia Article on Screws and Bolts for more information).
The carriage bolts are nice because they don't require a second wrench or pair of pliers to prevent the bolt from rotating during tightening. The square shoulder provides a locking surface to do this. It's a great feature until someone over torques the bolt and reams the area of wood that the shoulder embeds itself into out. This is where a torque washer is useful. We used them on the maze with great success.
They prevent the carriage bolt's shoulder from reaming out the wood because the torque washer has teeth that bite into the wood, like a nail does, and a square hole that mates with the carriage bolt's shoulder.
Using a wing nut, instead of a hex nut removes the need for tools to tighten the bolt and nut.
Another feature of the design that is worth noting is using over-sized holes for the bolts. This makes assembly of the maze easier, but increases the importance of using washers with nuts and bolts. I choose to use fender washers whenever possible because of the increased surface area spreading the force of the bolt head or nut out. In other words, the fender washer exerts less pressure on the frame, reducing deflection and deformation due to local stresses on the frame materials.
The hole pattern for the maze corner posts is a simple gird, based on the specs for the maze. Again, the holes are slightly over-sized to prevent problems from tolerance stack ups.
The walls of the maze are created a wall section and a corner post with a metal pin aligning the corner post to the hole pattern on maze floor. A wall and corner post are shown below.
In our build the corner posts were not painted; however, they should be painted to fully meet specs. The metal pin is a 1/8” diameter rod made of cold rolled steel. We spent a lot of time making the pins from 10' long sections of rod, because we thought it would be cheaper to make them versus buy them. In retrospect, buy the pins. They are sold as metal dowels. 1” to 1.5” long by 1/8” long should be fine.
Links showing .125” ⌀ round stock (1018 Alloy steel)
The round stock is cheaper than the dowels; however the tooling and time required to machine the round stock to size will completely offset any savings over the pre-made dowels. We needed 2 pairs of locking pliers (Vice-Grips), a bench grinder equipped with a grinding wheel and a wire brush, a metal cutting/abrasive chop saw and almost 16 man hours to complete the machining and de-burring processes.
The wall sections and the corner posts should be made with MDF.
The dimensions of the walls and corner posts are driven by the maze specs.
|Wall Section||15.6 cm (less the tongue)||1.2 cm||5 cm|
|Corner Post||1.2 cm||1.2 cm||5 cm|
The dowel pin is fit into a bore on the underside of the Corner Post. The bore should be .125”⌀ and 0.500” ± .05” deep and centered on each post (see photo below).
The reason for the centering of the bore is two fold.
This Corner Post was damaged by burrs on the machined pin. This is another reason why I strongly recommend using off-the-shelf dowel pins.
As the picture shows, there is fairly severe cracking around the hole. The cracks extend from the hole to the outside edges of the corner post as well as extending up the groves that receive the tongues of the walls.
This section covers how to construct the maze from the drawing package.
Being Written, current working copy (actively being written/edited) is below.
Based on our build, one of the key steps in making sure the maze meets or exceeds the specifications is selecting good materials. This means lumber that is straight and free of bows, checks, and similar defects. We did not use kiln dried lumber (KD), which in retrospect was a bad idea. KD lumber is ideal for this application because its moisture content is very low, which makes the lumber more predictable. If a KD 2×4 is good (straight, no bows, etc), then the odds are it will remain that way. Wet or green lumber has a very high moisture content (It can be wet to the touch) has a tendency to deform as the material dries out.
Careful design and construction techniques can mitigate the impact of using wet lumber; however, in the interests of minimizing waste and headache, using is KD lumber is strongly encouraged.
The maze is built in a modular fashion to allow the user to move the maze in a reasonable sized vehicle, store the maze in a reasonable sized space, and most importantly, move the maze through a normal sized door. Although there are doors that are over ten feet wide or tall, the door to the lab where our maze lives is not, so the ability to move the maze from where it was built to where it lives was a paramount motivation in the design of the maze.
Introducing modularity to the maze adds some headaches to the construction of the maze, namely that provisions for creating smooth surface that is nominally flat. The first version of the maze used a bunch of MDF sheets to smooth over the jumps in the maze deck. This solution yielded a very nice surface that, once painted, had nearly invisible seams. When we decided to put legs under the maze, we had to use magnets to find the screws used to secure the MDF. The second revision of the maze uses legs. The CAD rendering at the top of this article shows what the maze, with legs, looks like. The legs should be sized so that maze will fit though a standard door (our lab has an approximately 36” (91.44cm) wide door), 30” (76.20cm) is a good place to start. Our legs are are just under 30” long to fit some existing benches in our lab.
We attached the legs with a combination of 5/16” lag screws (also called lag bolts, but lag screws is generally considered the correct term). Braces are attached with course thread drywall screws. The bottom of the legs have a 5/16” leveling foot ordered from McMaster-Carr. The leveling feet allows the maze surface to be smoothed easily, without a huge amount of MDF and extra labor. This is critical if the maze cannot be left in a setup state forever.