News

Micromouse update 4, Makelangelo 7.3.4, Arm 3, and VR

Today’s monday update is about Micromouse 3, Makelangelo 7.3.4, Arm 3, and VR.

Micromouse update 3

In a previous post I told you about the track we’re building.

Micromoue track walls

The walls and floor are getting painted. We estimate the track is 85% done. In the photo above you can see the micromouse maze walls and pegs slot fitting together. We’ll be able to set up practice tracks at VHS before the event and the full track on the day of the event. Register your contest entry free right now!

Makelangelo 7.3.4

Over the weekend Makelangelo software has a small upgrade to 7.3.4. The “start”, “start at line #”, “top”, “left”, “right”, “bottom”, and “go home” buttons are now automatically disabled until you hit the home button. This is to prevent user error at the start – they’re not meaningful if the robot hasn’t already found it’s starting home position.

I’ve been cleaning stuff under the hood for the next 7.4.0 update. Until now converting a picture automatically places the picture in one place on the drawing robot. The improvements move everything to one coordinate system and one scale. This means that the code for each conversion style more concise because the style converter doesn’t need to translate between machine coordinate system and image coordinate system. It also opens the door to moving, scaling, flipping, and rotating images before converting to lines. Who knows? Maybe someone will add Photoshop style layers and an undo/redo system.

Here are some of the style conversion options, all using a picture of an owl’s eye.

Makelangelo 7.3.4 boxxy style

Makelangelo 7.3.4 zigzag style

Makelangelo 7.3.4 voronoi stippling style

Makelangelo 7.3.4 pulse style

Makelangelo 7.3.4 Sandy Noble style

Makelangelo 7.3.4 crosshatch style

All these styles are in the Makelangelo software already. You can download Makelangelo today.

Arm 3

My long term project is the next robot arm. Here’s an early draft, cutaway in Fusion360 to show the inside of the wrist.

early cutaway of arm 3

The goal is a ~100cm arm with 2kg lift and +/-0.5mm repeatable accuracy. No easy task! It also has to have all wires hidden inside so that it has a good ingress protection rating.  When it’s ready it will ship fully assembled, like all Marginally Clever robots, and it will be driven from Robot Overlord.

If you’re a long time follower then you know my last arm used two electric pistons, much like a backhoe you’d see at a construction site. Now I need some help with a math problem: how do I maximize the range of motion?

VR

Making robots easier to use is always high on my list of priorities. I want to program the Arm by pushing and pulling the real machine. But what if I don’t have the real machine in front of me yet? It would be pretty sweet if I could just push and pull a virtual model. It seems to me VR would be a natural fit. When I look at the choices out there, I’m being offered only awful choices.

  1. TOcculus Rift is owned by Facebook and (I hear) it likes to phone home all the time with mysterious updates and information sharing. On the plus side they have C/C++ developer libraries. In theory a Java wrapper can’t be far behind.
  2. The HTC Vive wants all their support handled by Valve and Steam, which means I’d need to register as a Steam game developer just to access their support forums. It feels like a lie! I can’t find anything about their developer libraries until I register. Edit: Tom G @ Valve just sent me a link to the OpenVR library, in C/C++. Java soon?  Edit: JMonkeyVR is the Java interface. Yessssss!
  3. The Microsoft Hololens looks ridiculous on the human head, and since when does MS support Java?

Final thoughts

You can get more up-to-the-second news from Marginally Clever if you follow me on Instagram.

News

Micromouse contest update 2

Monday again and I’ve got hot fresh tasty news about the micromouse contest.

Last week I posted about making the floor of the micromouse maze.  Since then we had some exciting moments figuring out how to make the walls and pegs.

Luke designed a jig that sits on the sled for the table saw and makes it easy to get the exact groove cut needed for the pegs.  We tried to make the pegs from 1/2″ birch ply but the material was too brittle and chipped to shreds.   In the video above you can see we’ve switched to MDF.

To keep manufacturing simple we used the exact same groove on the wall pieces.  The gap between each peg and wall is 3mm thick, 5cm tall, and 6mm wide.  As it turns out, a laser cut piece of acrylic is a perfect fit.  The acrylic comes with a protective layer on both sides.  The layer adds just enough material to make the fit snug so we won’t need any glue.  Excellent!

How many micromouse maze pegs?

Having a system to make walls and pegs is great.  How many do we actually need to make? The rules say there are 16×16 rooms with a wall around the outside edge. That means there are 17 * 17 = 289 pegs. What about walls?

Absolute maximum

Well, there are 16 wall segments to go across and 16+1 rows of walls, so that’s 16×17 for just the horizontal walls. times that by two to get the maximum number of walls, that’s 16x17x2 = 544.

Realistic maximum

In reality there won’t need to be so many walls. Every room has to be reachable from at least one other room. The easiest way to draw this would be a snaking S shape.

Now it’s easy to count them out. There are 16×4 wall pieces around the outside, and 15×15 pieces inside the maze. That’s 289 wall segments. Funny coincidence: that’s the same number of pegs!

Final thoughts

Next step is to paint everything regulation colors and get a timing system to record each race.

A big hello to everyone from Makerfair joining us for the first time. Hello!

Follow me on instagram if you want a to see the acylic pieces as they’re being made.

Tutorials

Introduction to LCDs

An LCD is a great way to communicate a lot of information fast from an Arduino. Combine it with a few buttons, and it suddenly becomes a great control interface for your new machine, whatever it may be. However, LCDs only use 16 pins to communicate with other circuit boards, but the Arduino UNO only has 13 pins. We’re going to need to use something in between to let us do more with less. This is where an I2C backpack comes to the rescue. The only tricky part is that each I2C backpack and LCD are slightly different and require some small tweaks. Today, let’s go through it and show you how the LCD/I2C combo works out.

Check out our final product:

(more…)

News

Robot Arms Running Marginally Clever Open Source Code

Marginally Clever open source code is gaining traction around the world as young makers use it to build ever more amazing machines.

Pete from Fablab013 in the Netherlands says,

we used your gcodeCNCdemo to power our 3d printed robot arm.

Print time : 80+ hours
programming time : 10 minutes

Next thing you know they’ll upgrade to Evil Minion firmware so they can drive it from Robot Overlord – and we’re looking forward to it!

As we’ve said for years, affordable robot arms are the next step.  Marginally Clever wants to make that future get here quicker.  Part of that plan is we’ve been giving away the code to make it happen.  We’ve also designed our own arms, and we’re looking to help other talented robot arm designers.  We want to work with anyone else who dreams like we do.

In the News

Micromouse contest update 1

An update on the 2016 micromouse contest coming to Vancouver Mini Maker Faire June 11-12. The contest poster, tips, and details on the maze.

First, the poster. Please share with everybody!

micromouse 2016 contest poster

Second, here’s a great way to get around the 2s penalty for touching the robot when it’s in the maze.

Third, the maze itself is coming together. We’re having great fun building a maze that’s interchangeable and portable. This contest has been run in various places around the world for 30 years and yet no one’s published a satisfying tutorial how to build the track.

Special thanks to MicrmouseUSA.com for the photo of their robot.