Robot drive sub-systems

There’s not been much discussion this year about the best style of drive for a Skyrise robot, not really surprising as we don’t have the obstacles (bump, sacks) to deal with that we have for the last couple. I’ve been thinking about building a new drive to be able to look into programming the accelerometer and gyro as there seems to be renewed interest in these sensors. The last robot I built had a small chassis (open source robot) and this time I wanted to build something with the full 18x18 dimensions. I also decided to try and incorporate some features that my students never seem to think about, these were my priorities.

  1. Protected motors, often motors are placed directly driving a wheel and end up on the inside of the robot. They tend to get in the way of game objects and can end up being damaged.

  2. Easy wheel installation. We always seem to be trying to get that last washer onto the axle using tweezers. I wanted to be able to easily install wheels.

  3. Ability to quickly swap between mecanum and omni wheels.

  4. Easy access to tighten the motor screws, easy to replace a motor.

  5. Potential for different gear ratios without a complete redesign.

  6. The ability to use IME’s or quad encoders on all four wheels.

  7. Screw heads on the outside, nuts hidden on the inside.

So with these goals in mind I came up with the following design.

Left and right sides are completely symmetrical, there’s no need to mirror image during the build. Each side is based around a single 1x5x1x35 C channel that acts as a sort of mid rib. To the C channel connect standoffs to both sides, the general construction of the mid rib is like this.

You can see that wheels are installed from the outside, the 1:1 gearing (using three 36 tooth HS gears) is on the inside. A 5x10 plate covers each wheel, a 15x25 plate covers the motor section, the drive can just accommodate a mecum wheel and is 4 inches from the outside to the inside of the gear housing. This shows how an individual wheel is replaced.

Using the 36 tooth gear allows the motor screws to be accessed for tightening as there are holes in the gear that align with holes in the structure. There’s enough room to use motors with or without IMEs installed and replace them without any disassembly of the chassis.

Gears can be replaced without removing the wheels, just four screws are used to secure the 1x5x1x10 C channel on the inside of the chassis. Other gear ratios (for example, 60 tooth on the wheel giving a 3:5 ratio) need the addition of another 1/8 inch spacer to the gear housing and movement of standoffs, I nearly always run 1:1 (or 1.6:1 using internal motor gears) so not really a concern.

This shows access to the traditionally difficult to access inner motor screw.

To connect the left and right parts of the drive a 1x5x1x25 C channel was used on the top and a 5x25 plate underneath. These two were connected with several standoffs and some left over 1x5x1 C channel after cutting the sections for the gear housings. This create a box like section the ties the two halfs together, I had expected some flex in the chassis but it’s actually very stiff.

And here is the final assembly.

I didn’t have enough 36 tooth gears so used 12 tooth sprockets on the front wheels as an alternative, only minor changes to the placement of the bearing flats was needed to accommodate theses. Two of the motors are encoded, I also fitted a quad encoder on one back wheel as an example. I added a further 1x5x1x25 C channel to support the battery and cortex.

So there it is, my new test bed drive. It would have been pretty useless for the last two competitions but may make a good platform for a skyrise robot. The robot measures 17.125 (width)x17.5 inches, there’s about 9 inches clear between the gear housings so enough room for a skyrise cube (although team 8888 has other plans there), weight with cortex about 7lb but a heavier drive may be a good thing this year.

There are a few more CAD drawings here.

Please don’t quote this post as the images are linked.


Lots of interesting ideas. I’ll have the kids take a look at this. Thanks for sharing. :slight_smile:

Very nice! I really like this approach. I’ve never tied putting the motors on the inside, but this design shows how practical doing it that way can be.

Is this all aluminum?

1 Like

Yes, all aluminum. I need to analyze the design and see where the major contribution to the weight is (just looked up mecanum wheels, they add 1.6lb), I was hoping the chassis would weigh around 5lb, so it’s almost 40% above where I would have liked to be. I tend to over engineer my designs a little even though they are just for my own amusement and are never used in competitions.

1 Like

How much do all of the standoffs weigh? Looking at the CAD, there seems to be a lot of them.

Ok, so using weights from the VEX web site and also from Inventor when they were not available, here is the approximate breakdown (in pounds).

               unit weight  qty total weight
wheels              0.410     4       1.640
motors              0.192     4       0.768
C channel 5x25      0.211     3       0.633
C channel 5x35      0.298     2       0.596
plate 5x25          0.146     3       0.438
C channel 5x10      0.090     4       0.360
standoff 2"         0.010    33       0.330
cortex              0.302     1       0.302
gears               0.025    12       0.300
screw 0.5           0.004    73       0.292
nuts                0.004    64       0.256
screw 1.25          0.009    28       0.252
screw 0.375         0.004    44       0.176
shaft 4"            0.018     8       0.144
misc (spacers etc.) 0.126     1       0.126
plate 5x10          0.060     2       0.120
bearing flat        0.003    28       0.084
screw 0.75          0.005    16       0.080
screw 1.0           0.007     9       0.063
standoff 1"         0.005     8       0.040

and a simplified version

wheels      1.640
motors      0.768
structure   2.147
screws      1.119
gears       0.300
cortex      0.302
misc        0.724

There’s really not much weight reduction to be done without compromising the structural integrity. Aluminum screws would perhaps help, a few less standoffs, cut away some of the C channel, perhaps save 1 lb. The basics (wheels, motors, cortex, gears) add up to 3 lb alone.

Swapping 4" omni wheels for the mecanums would save almost 0.75 lb.


Why would you want to reduce weight on the base? On a skyrise robot, which would need to extend over 60" tall, needs a heavier base in order to have a low enough center of gravity to stay upright reliably.

I like this design though, very clean and versatile.

Nice design, James. The only change I would make is to have the 5x bridge deck extend to the outermost plates. It would add a lot to the stiffness of the drive modules in the vertical plane without meaningfully changing the weight. I’m a little nervous about the ability of standoffs to remain perpendicular to their mounting plane once weight of the game mechanism and objects is added.

I played with a similar idea for modular drive systems in 2009, and combined it with an idea I had for a swing-shifted vertically-oriented 2-speed transmission. The transmission didn’t work particularly well (it takes a lot of force to push gears together), so I never took it any further. I did place the motors on the outside of the chassis based on the same idea you had about clearing out the center, but I used cantilevered wheels to protect the motors. The modular drive train appeared again in Exothermic Robotics’ Clean Sweep robots in a different form (where the frame rails were traditional, but the 6-motor drive system was attached with four screws and two lock collars).

Here’s a picture:

1 Like

It was just an exercise to understand where the weight is, everything is a compromise, a four motor drive still has to move the total weight of the robot and it may be better to be able to use that on the other sub systems.

That could be done, I wanted to leave easy access to the motors. A custom bracket would be needed to secure the outer plate to a longer C channel.

I was concerned about this as well, but it feels really strong, I’m not able to distort anything by pushing down or flexing the chassis. The 1/4" spacers help somewhat as they sit on the edge of the C channel and reduce the total distance that the standoff has to support (see pic below). I was going to add diagonal support to stiffen things up, this was actually the third or fourth iteration of the design and it just didn’t seem to need it. I don’t plan on building a complete robot (although you never know) and the lift system would almost certainly mean modification to this design. I also wanted to leave a little room for students to improve on this, I used lots of standoffs because that what I had in the parts bin and didn’t want to cut too much aluminum, there are other ways to achieve the same end result.

1 Like

Hi jpearman, I would just like to say awesome design! after my team put quads onto our wheels we could never get to the motor screws without taking apart the base (that’s going to change this year after seeing this :slight_smile: ).

I hope I’m not missing anything obvious here, but is there a reason when you actually built the drive one of the c-channels connecting both sides of the drive is flipped around (relative to the CAD picture)? I see a battery and cortex attached to it, but that could be achieved even if it was flipped around.

1 Like

Probably, at the time it just made a more stable platform to mount the cortex. I have wires running underneath the cortex in the C-channel shown in the CAD. The second C-channel was not in the CAD and I just quickly added that to test everything out.

1 Like

That piece C-channel in the center of your chassis caught my eye. Mind if I borrow that little tidbit of structural design from you?

No problem.

I was just looking at these features and I realized, how would you really use features 2, 3, 4, and 5 at worlds? If you’re at worlds than you probably have your design nailed down. Also, to create the robot with the least amount of friction, you would have to direct drive the motors. I would agree that we need to protect the motors, but I think there are ways to protect the motors and still keep a direct drive. I can see the benefit in the beginning of the season or for testing, but who wants to change their drive right before Worlds? I am not trying to be mean or anything. I highly respect jpearman for all he has done for this community. I just figure this is not going to be extremely helpful long term. Of course, I could be very wrong on this. It’s just my two cents.

A simple answer would be to just say that this robot is not designed to compete at worlds, however, I still think these features have merit.

Wheels get broken. I hear some teams swap wheels for skill runs. Axle’s get bent.

This is always needed, worlds or not. One solution to overheated motors would be to swap motors between matches.

Agreed, by worlds this should have been decided upon, however, we have heard in the past that robots that worked well back at school or in local tournaments have had issues at major competitions and tripped the PTCs in the drive motors. One theory is that the use of anti-static spray on the fields increased friction, another is that drivers are more aggressive in a competition situation. A solution to this can be changing the gear ratio and slowing down the robot.

My real reason for all the above is that this is a test platform. I want to be able to experiment without having to rebuild. I wanted to be able to compare omni and mecanum wheels when under PID control, which ones are better for autonomous etc. Not had time to do any of that yet, I got distracted by the lift design and then the day job became busy, Oh well.

I also throw my designs out to get discussions going :slight_smile: I like to hear that you disagree.


I see your point. I can see the usefulness a bit more now. Now that I think of it, because of its flexibility of changing different speeds and different wheels, I can see that this would be quite useful for testing the field in the early season (April-September). For instance, you could find out what speed is optimal or whether holonomic movement is important this year just by driving it around for a while. Just a thought.

1 Like

We built a design like this for testing as well. Works awesome to move motors inside and free up open space for cubes to slide in between the drive so we. Can use a passive intake.

How do you have your robot “pull apart” in pictures 2 and 3 of your original post? Sorry this is off topic, but I would like to do something similar.

Thanks, and I love the drive :slight_smile:

1 Like

Not sure if this is how he did it, but this is one way

1 Like

Autodesk Inventor Publisher

You can import the assembly from Inventor and then create documentation and/or animated exploded views. Free for educational use.

1 Like