Let's Make Robots!

Tri-wheel stair climber


It seems to me that for a robot to be really useful in the home or workplace then it must be able to handle stairs. With this in mind I want to design a stair climbing robot chassis. This robot is my first experiment with Tri-wheels.

The chassis is just 4x Wild Thumper 75:1 motors driven by a Wild Thumper controller in a clear acrylic base. As this is only an experimental platform for the wheels I am controlling it with a TV remote and an IR receiver.

This first set of Tri-wheels is made using a new 80mm diameter wheel that has been produced by DAGU. The gears are steel to handle the high torque required when climbing.

The clear triangular frame is made from 3mm thick milled polycarbonate. I used CNC milled polycarbonate instead of laser cut acrylic because it is tougher. The acrylic is fine for the chassis but too brittle for the wheels. All shafts are ball raced to reduce friction.

You can see here that the chassis consist of 4 ribs, a top plate and a bottom plate. They are all held together by 56 small steel angle brackets and 112 small M2x5 screws that give the chassis good strength and rigidity. In hind sight I could have used less but this is just a prototype.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Man, that Twister design is very good! The Mars Rover guys at JPL should've had a look at that. ; j

You will probably find that they avoid using a triwheel design because of the weight. Each of my wheels has 7 steel gears and 13 steel ball bearings that make it quite heavy. Even if they use Iron Mans gold/titanium alloy thats still a lot of extra parts to carry plus you then have more things to go wrong.

 

Yes my first attempt was a bit rushed.

As I was only testing the wheels I just whipped up a very simple chassis without considering center clearance. I have already designed a new chassis to solve this problem so I can test the wheels and determine the maximum size stair they can climb.

The Twister was a great example of the tri-wheels. Lego has definitely come a long way since I was a kid.

This is a very tricky task but it looks like you're off to a good start OB, best of luck with the development =)

That is an awesome first run. Just a thought, you are legues ahead of me, but would a mechanism to lock the wheel gear and center gear together on demand be enough to get the desired results when climbing the stair face? I'm definately going to keep an eye on this one. Great idea.

In this case, if I could have caused the rear geartrains to lock (since gravity didn't) then the extra height of the back tri-frames being on their point would have probably been enough to get the chassis over the step.

There are two problems involved with that idea.

  • First of all designing a suitable mechanical brake system. Perhaps a solenoid with power supplied through sliprings?
  • Second, having an autonomous robot know when it is necessary to use such a system.

Tri-wheels are a differential drive system that automatically switch between driving the small wheels and rotaing the mechanism depending on which requires the least torque. For the sake of a better chassis design or larger wheels I would prefer to leave it automatic K.I.S.S.

I saw something, and I think you saw it too. Looks to me like your bot is in harmony with the stairs. Both sets of wheels are hitting the stair riser at the same time so the front wheels can't help pull the bot up to the next step. I think this is why you attempted to climb the steps at an angle, effectively "shortening it's frequency"? Maybe I'm just crazy. -_-

 

I did not measure the stair tread. Since the tread and riser height can vary from house to house my aim is to make a chassis that can handle any step. That said, stairs that just have treads with an air gap between do not work with Tri-wheels. Tank treads are required for those kind of stairs.

I should have said that  the front and back sets are hitting the stairs at the same time, to be more clear. Maybe it is irrelevent, but maybe it is better if the bot is only trying to climb one step at a time, if that makes more sense.

I think I'm getting the picture, so basically it uses the weight of the bot and traction to "lock" the gears.... as long as the small wheels can get enough traction to transfer the torque to make the whole mechanism spin, it doesn't need any of that Rube Goldberg nonsense :). Then it turns into a "simple" matter of weight quantity and placement, and wheel size. You are right again, that is so much easier :p This is the first time seeing this mechanism, so it intrigues me. Keep up the awesome work.