Two-Speed Chassis

This was designed for SWAT Robotics (824) here in Seattle, WA. While they haven’t definitely decided against this drivetrain and chassis, they’ve indicated a strong preference for the use of 80/20 and the ease of construction it provides. So, to be sure that this design doesn’t go to waste, I’m sharing it here.


This design exists as an alternative method of achieving two-speed capability without employing a shifting transmission. Instead, it will rotate two sets of differing diameter wheels into and out of place along the carpet. It is equipped with two, two-motor, single-speed gearboxes and shifts via two, 1.5" bore, 7" stroke pneumatic pistons.

Low-Speed Configuration
To meet size constraints, the chassis begins each match in the low-speed configuration. This also ensures that the robot will be formidable in pushing matches should the shifting mechanism fail to operate. Top speed is approximately 4 ft/s.

High-Speed Configuration
The high-speed configuration offers a longer wheelbase for increased stability when rapidly accelerating and decelerating. Four omniwheels allow for zero-radius turning capability and low friction in all turns. Larger wheel diameter and additional gearing allow for a top speed of approximately 10 ft/s.


  • Increased control over design considerations – By switching wheels and changing ratios through differing wheel diameters rather than by simply changing ratios in a transmission, this design offers more control over the chassis’ high- and low-speed capabilities. Varied wheels types, sizes, widths and tread materials can be used on the high- and low-speed wheels to achieve precisely desired operating characteristics that best suit the game.
  • Offers increased stability for high-speed translation and decreased resistance to turning in low-speed – this allows for it to rapidly change direction at high-speed with reduced chances of instability or tipping over. It also allows for more torque to be exerted onto other objects during turning.
  • Less high-precision machining – Though, as shown, good machining capabilities are helpful, it is possible to employ this solution’s concepts with less machining ability than that traditionally associated with most shifting transmissions. This isn’t absolutely true, however.
  • Stair climbing – By switching from low- to high-speed configuration, chassis can climb ledge approximately 5" in height.


  • Weight – heavier than traditional shifting transmissions.
  • Size – takes up more space than traditional shifting transmissions.
  • Complexity – more moving parts.

This was created with Solidworks 2003. The project files, in whole or in part, are available to anyone who asks. Feedback, constructive criticism and questions are encouraged.

I just looked at this design, and my initial reaction is simply “wow”. This is awesome, Krass. I especially like the change in traction and wheel characteristics between high and low gear.

Simply put, this is a sweet design.

Andy B.

M., this looks pretty cool, but I’d be interested in seeing your “simplified” design that you allude to for a couple of reasons.

First of all, any team that can make this could go above and beyond with a shifting transmission. I spot a four motor drive train, meaning they have to be precise enough to make the gearboxes and the gear train. Those wings are pretty massive, and would require a lot of patient machining on a sizeable mill. Making omniwheels like that would be some serious machining in and of itself. By looking at this design I think the “no high precision machining” advantage is moot.

I also take issue with the claimed curb-climbing capability. The front and rear omniwheels drop at the same time, which means you could get on the curb, but then what? It looks like the robot would high-center before it got the rest of it’s frame over the curb. If you could show me a step-by-step demonstration of the process, I’d appreciate it.

My main concern is the limited amount of space for subsystems. Any large arms or grabbers would be impeded by the middle bar spanning the length of the chassis. I was also going to say that two pneumatic pistons are used to shift drive trains, but I guess that’s a moot point because conventional trannies tend to use pistons anyway. A pneumatic question that persists, though, is what happens if a subsystem requires a pneumatic piston? Shifting speeds with massive two 7", 1.5" bore cylinders would drain the storage tanks pretty quickly.

One thing I do like, though, that you didn’t mention is that in high gear (if the pistons are kept to a lower pressure) you almost seem to have an independent suspension, allowing individual wheels to bounce around a little bit and dampen any bumps in the road. This could be useful if FIRST decides to give us a few off-roady type environments (here’s hoping, that’d be sweet :slight_smile: )

I’m also working on a frame-changing transmission for this season. I’m not going to make it public before build season, though, because there’s a good chance my team will use it. I’d like to get in touch with you and talk about it though :slight_smile:

PS: I’m pretty good at Inventor (I worked as a draftsman over the summer for a bit, had some fun) but that’s the only real solid modeler I’ve tried. How easy is it to switch between Inventor and SolidWorks?

If a team were to put tread material on the “low speed” wheels, similar to the design from which this, um, monster (for lack of better word :p) came from.

If you did that, it’d enable you (I think) to mount the stair, and then the tread material would pull the robot up as you powered the motors forward. Once you’re up, pop back up into high speed, and you’re good to go.

Similarly, to get down, you could drive the low speed wheels up to the ledge in low speed (obviously would require a high amount of driver control, or some sort of feedback system), pop “up” into high speed, but since the high speed front wheels are over the air, they fall back down, hit the ground, and you can smoothly transition back down to the lower area (or much more so that just driving off and hoping you land without breaking anything).

The only problem I can see here is, well, Krass’s designs do tend to be a bit more complicated to build than they look on paper (a problem for teams with less materials, but well within range of any midsized team who has a decent supply of robot-building parts). But, if you can get it working, it’s usually a pretty formidable robot. And not to discourage newer teams (is 810 still considered a newer team?), but this is a design that you may want to hold off a bit on, at least until you’re sure of your capabilities, because yes, it will perform, but it will take you a long time to build this.

Otherwise, I’m glad to see that “El Toro” may now have a distant cousin/decendent/etc, and I really hope a team does go with this design this year, because it is quite an amazing looking robot.

Keep up the good work Krass :).

Beautiful… I love robots that have different sets of wheels to shift gears. But someone has to be a skeptic… how much does it weigh? We tried a similar thing (with four motors and two sets of four wheels) in 2002 but had to take off two wheels, two motors, and a whole component from our robot to be able to get underweight. I guess if we had designed with weight in mind from the beginning we would have had much better success though (a subtle hint to any young impressionable team members out there!!). Your design looks much more weight efficient. Did you do mass analysis with Solidworks? Hopefully I’ll somehow get to see the finished bot in person (or at least you can post some pics at the end of build season)!

Good luck

  • Patrick

question… how much torque would you have when you are driving in low speed configuration…and how long would it take you to get to high speed configuration from low speed…??? should i guess about probably a sec or two… since you are using pneumatics to switch…?

the design looks awesome…

I love it as a drive train! its a great idea, Seems like you would be cramped for space and/or weight but still a cool idea.

There 101 different directions you could go with an idea like that to fit your own individual robot each year.

Keep up the good work.

I saw a robot that did this at the SoCal regional last year, but I do not recall the team number.

What is the height of the crossmember when raised? How much is the base raised when in highspeed mode? It appears from the images that it isn’t that high with plenty of room around it to place things.

However, it does appear to severly limit placement of manipulators low to the ground, which has traditionaly been part of the game in some way.



Hurray for shared innovation.

That would be 980.
Saw them at IRI.
Heck of a bot.

In terms of actually transferring power to the two sets of drive wheels, are you going to be driving the low-speed wheels directly from the gearboxes, and then running chains to the high speed wheels? (No driveshafts appear in the drawings.)
Are you going to have enough room to run sprockets and chain (with any necessary ratio) inside the pivoting triangular frames (the red parts with wheels attached)?
Also, both sets–front and back–of low-speed wheels need to be powered. Does this imply that you’ll be running a drive chain inside of the C-channel, from shaft to shaft? If so, it looks like you’ll be limited to relatively small sprockets (max 2" to 2.5"), which, as we found out last year, can eat chain, given a powerful drive system…
With what appears to be 1/8" plate in the centre of the chassis, are you concerned at all about flexing/buckling (due to collisions, or the 200-lb pneumatic fury of those pistons)? A crossmember of some sort would surely minimize this. (Maybe that was going to be added later, in the final version of the design?)
Lastly, will those 1/2" rods that support the entire wheel assembly on both sides be enough to handle the stresses of driving? If the field is flat, you might be just fine, but with something like last year’s ramp, they may prove insufficient for the bumpy ride that the robot might endure. (The bolt holes in the C-channel where these attach are going to be rather stressed if the robot falls even a few inches.)
OK, that’s enough constructive criticism for now…otherwise, though, it is really quite an interesting design. (And very nicely executed.)

I appreciate everyone’s feedback so far. I promise to answer all the questions you’ve asked as soon as I can. I think I need to do some math again since I can’t find the calculations I did for output speed and output torque, particularly.


Nice design.

Might I suggest that two of the omniwheels (on one end) be changed to regular high friction material. This will give you much more control. With four omniwheels, it can slide around too easily, especially if it gets bumped.