Sawppy the Rover

Motorized model of Mars rovers Curiosity and Perseverance For Under $500

Similar projects worth following
Sawppy is a motorized model whose layout and proportion mimics that of Mars rovers Curiosity and Perseverance. It faithfully reproduces the suspension kinematics of real rovers and is intended to be a hardware platform for future software projects in autonomous operation. Go forth and boldly explore the back yard, Sawppy!

Sawppy the Rover was inspired by JPL's Open Source Rover project. Most of the differences between Sawppy and its JPL inspiration were motivated by a desire to reduce cost and complexity. JPL's rover is designed for education, to be assembled by a school team and give a robust foundation for structured curriculum. Sawppy is more suited for individual hobbyists like myself who are happy to tinker and willing to make some trade-offs to lower cost.

The budget was $500, and getting there required the following changes:

These two major design goals can be summarized as: Servo Actuated Wheels, Printed Interconnect For Extrusion. The acronym SAWPIFE led to the nickname "Sawppy".

See the "Links" section for pointers to additional information:

  • Live Onshape CAD file: This is where I'm tweaking and building Sawppy in full public view. Be warned the live file has upsides (latest ideas!) and downsides (latest idea doesn't work!)
  • Github: This is where the assembly instructions currently live. It also has a snapshot of Sawppy components in STL file format. These parts may lag behind the live CAD data, but they have been printed, installed, and proven to work on my rover.
  • Build Blog: The history of Sawppy, including stories of design goals and lessons learned from failures.

Since the time I declared Sawppy version 1.0 (mechanical foundation) complete and posted assembly instructions, others have built on top of what I've released to the world! They're outlined in the project log "Sawppy Builders" entries, page 1 and page 2 with more to come as I hear about more rovers.

Sawppy V1 Schematic.png

Bare bones Sawppy version 1 schematic

Portable Network Graphics (PNG) - 192.78 kB - 03/13/2019 at 06:14


  • 3 × 1 kg spool of 3D printer filament Exact amount of filament used will depend on slicer settings, see "Instructions" section.
  • 10 × LewanSoul LX-16A Serial Bus Servo Design also can be adapted to use Dynamixel AX-12A or Herkulex DRS-0101
  • 3 × Box of 100 M3 * 8mm hex bolts Fastening 3D-printed components to aluminum extrusions
  • 3 × Box of 100 M3 hex nuts Fastening 3D-printed components to aluminum extrusions
  • 3 × Box of 100 M3 washers (< 10mm diameter) Fastening 3D-printed components to aluminum extrusions. Design can accommodate up to 10mm diameter washers, larger washers will require modifying CAD file.

View all 17 components

  • Rhys Mainwaring’s ROS Melodic Software and Simulator for Sawppy

    Roger01/31/2020 at 22:37 3 comments

    When I created Sawppy, my first goal was to deliver something that could be fun for robotics enthusiasts to play with. The target demographics were high school students and up, which meant creating a software stack that is self-contained and focused enough to be easy to learn and modify.

    To cater to Sawppy builders with ambition for more, one of the future to-do list was to write the necessary modules to drive Sawppy via open source Robot Operating System. (ROS) It is a platform with far more capability, with access to modules created by robotics researchers, but not easy for robotics beginners to pick up. I’ve played with ROS on-and-off since then, never quite reaching the level of proficiency I needed to make it happen.

    So I was very excited to learn of Rhys Mainwaring’s Curio rover. Curio is a Sawppy sibling with largely the same body but running a completely different software stack built on ROS Melodic. Browsing the Curio code repository, I saw far more than just a set of nodes to run a the physical rover, it includes two significant contributions towards a smarter rover.

    Curio Rover in Simulation
    There’s a common problem with intelligent robotics research today: evolving machine learning algorithms require many iterations and it would take far too long to run them on physical robots. Even more so here because, true to their real-life counterparts, Sawppy and siblings are slow. Rhys has taken Sawppy’s CAD data and translated physical forms and all joint kinematics to the Gazebo robot simulator used by ROS researchers. Now it is possible to work on intelligent rovers in the virtual world before adapting lessons to the real world.

    Rover Odometry
    One of the challenges I recognized (but didn’t know how to solve) was calculating rover wheel odometry. The LX-16A servos used on Sawppy could return wheel position, but only within an approximately 240 degree arc out of the entire 360 degrees circle. Outside of that range, the position data is noisy and unreliable.

    Rhys has managed to overcome this problem with an encoder filter that learned to recognize when the servo position data is unreliable. This forms the basis of a system to calculate odometry that works well with existing hardware and can be even faster with an additional Arduino.

    ROS Software Stack For Sawppy
    Several people have asked me for ROS software for Sawppy, and I’m glad Rhys stepped up to the challenge and contributed this work back to the community. I encourage all the Sawppy builders who wanted ROS to look over Rhys’ work and contribute if it is within your skills to do so. As a ROS beginner myself, I will be alongside you, learning from this project and trying to run it on my own rover.

    (Cross-posted to

  • Sawppy Builders (Page 2)

    Roger08/23/2019 at 07:48 1 comment

    This is the second page of Sawppy Builders, for Page 1 click here.

    Steve (jetdillo)

    I met Steve at Maker Faire Bay Area 2019 in May, where he showed me pictures of a few of his Sawppy parts. It is now August 2019 and his Sawppy is mechanically complete. Steve is a HomeBrew Robotics member like Marco, and the new rover is neon green to distinguish from Marco's neon yellow Sawppy.

    In terms of modifications, Steve is adopting Marco's faster wheel drive motors, and instead of cutting drive shafts from metal Steve has printed his from NylonX. It'll be interesting to see if that material is strong enough for the job!

    See more details at Steve's YASP (Yet Another Sawppy Project) page.

    Leif Sorgule

    Leif is teaching a robotics course for high school students and he thought Sawppy might be an interesting team project for the class. The class will start with the Arduino control option, given cost and learning resources available for Arduino. It'll be fun to see what the students dream up for their own Sawppy modifications.

    Patrick Leiser

    Patrick's Sawppy variant (named Swappy) made an appearance at the Rocklin Mini Maker Faire, representing the Sierra College Robotics Club. It might be the first of Sawppy's siblings to be chauffeured around in a Tesla Model 3, though this picture shows it doesn't quite fit in the frunk.

    TeamSG Aussie Sawppy

    Sawppy has a brother down under! This is a father-and-son team putting a rover together, and their chosen yellow-and-black color scheme looks very sharp.

    ... and hopefully more rovers to come!

  • Sawppy at Maker Faire Bay Area 2019

    Roger07/28/2019 at 23:53 0 comments

    I took some footage of Sawppy's adventures at Maker Faire Bay Area 2019 and assembled them into three videos.

    It was raining at the event, so Sawppy needed to put on a raincoat before venturing outdoors:

    It was the first time Sawppy attended a three-day long event and one of the steering servos failed on the third day.

    After the event, Sawppy needed some cleaning and minor repairs.

  • Curiosity Rover 3D Resources

    Roger06/25/2019 at 20:52 0 comments

    Prompted by a question on the JPL Open Source Rover web forum, I compiled all the 3D resources I had collected on Mars rover Curiosity. This reference data helped Sawppy match Curiosity's overall proportions and suspension geometry, which was my goal of making a mechanically faithful motorized model. I stopped there, but others rover builders like @lmckeega are working to improve accuracy in appearance so I thought I'd share these resources to assist such efforts.

    3D web sites

    My starting point was JPL's official open source rover website whose opening animation has a 3D model of Curiosity cruising on a simulated Mars surface. I tried to extract that 3D mesh without success.

    On a similar front, we could see a 3D model of Curiosity in the "Experience Curiosity" website. It's possible this is using the exact same data set as the OSR, but still I'm not enough of a web developer to pull out the 3D data.

    Finally we have a 3D model visible on Curiosity's mission site. Again it may or may not be the exact same one used in above two sites, but the difference here is that we have a "Download" button. Clicking that button results in a file named Curiosity_static.glb. My laptop running Windows 10 has a 3D Viewer app installed by default, which was able to view this file. I don't know what viewer software would be required for other platforms.

    3D printing

    A web search for "Curiosity 3D Model" and similar keywords would repeatedly lead me to a 3D-printable static model. Unfortunately, for my purposes this model is not useful. The geometry of this model were modified to be friendly to 3D printing and is not a faithful representation of Curiosity.

    3D animation

    However, on the same site, there are two Curiosity models for the free open source 3D animation program Blender. As far as I can tell, these two models have the same 3D data but with different textures. "Clean" is factory fresh Curiosity, and "Dirty" represents Curiosity after cruising on Mars for a while.

    The advantage of these files is that suspension parts are separate elements that can be animated to show suspension articulation. I believe these files formed the basis for Gazebo simulation described here.

    It also means we can split parts apart for a closer look. However, this file only has enough detail for animated graphics, it does not have enough detail for CNC machining. Much of the surface detail are represented by bitmap textures instead of 3D mesh data.

    While there is not enough detail for building a high fidelity model, these files were the best resource I had to measure component sizes and their relative layouts. I was able to bring them up in Blender, switch to orthographic projection view, and get images of Curiosity free of perspective distortion. In case that's useful to anyone, and you don't want to install & run Blender just to obtain those images, here they are:

    (Cross-posted to

  • Wired Joystick Controller With Arduino

    Roger05/24/2019 at 00:37 2 comments

    As a presenter it was great to see the wealth of information provided by Maker Faire Bay Area organizers. It is clearly the compilation of many years of experience and lessons learned. One item caught my eye: anyone using wireless communication should have a contingency plan in case wireless fails.

    I've definitely encountered this before with SGVHAK Rover and Sawppy on the standard 2.4GHz WiFi band. So far I've been able to avoid most problems by adding a dual-band router and moving up to the 5 GHz band, but in a large event like the flagship Maker Faire, even that might not be enough.

    So as part of my Maker Faire prep work, I designed and built a wired joystick controller with an Arduino as the microprocessor. When active, it replaces the Raspberry Pi and wireless router on board Sawppy. This also serves as a lower-cost alternative to rover builders or those who like the thought of walking their rovers on a leash.

    More resources:

  • Sawppy Will Be At Maker Faire Bay Area 2019

    Roger05/10/2019 at 19:43 0 comments

    I know I've said I wanted to keep my project page focused on rover design and evolution and less on rover adventures in the wild... but I'm going to break my own rule for this one. I'm super excited to have been accepted as a maker presenting their project at Maker Faire Bay Area 2019. It's not just A maker faire... it is THE maker faire. Most of my travel logistics have been nailed down, it is happening!

  • Sawppy Builders (Page 1)

    Roger05/07/2019 at 23:03 0 comments

    I started my Sawppy project in March 2018. By May 2018 I had my first rolling chassis but it was fragile. Every time my rover broke, I learn a little more about mechanical design, and Sawppy improved over the following months. I also started writing assembly instructions and supporting documentation to help any other interested people build their own Sawppy, not knowing if anyone would take me up on my offer. It was extremely gratifying to see other people have indeed accepted my invitation!

    This post recognizes those who have embarked on their own Sawppy adventures, roughly in the order of when I learned about their efforts. Sometimes I learn about their ambitions before they got started, sometimes I learn about it only after their rover had been completed. Given this, it is likely there are other Sawppy builders out there I don't know about at all! But that's fine, I just love the fact there are more Sawppy rovers running around.

    I used to list a few Sawppy builders in my main project description page, but the list has grown too long to fit in that space. I'm going to track Sawppy builds on this project log entry, editing it as I go to add more rovers as they come online. I don't know if there'll ever be a day when even this is too unwieldy to track all the Sawppy builds out there... but as far as problems go, that's one I would be very happy to have.

    To everyone who decided my project was worth your own time and money and effort to build: Thank you.


    Daniel Perron

    Daniel was one of the first to jump in and start exploring the information I had released. Daniel also helped spread the word of Sawppy via comments on various web page comments about JPL's Open Source Rover, and for that I'm grateful.

    I designed Sawppy to use Misumi aluminum extrusions because I saw Misumi had distributors worldwide. Unfortunately my hope of easy availability turned out to be wrong in Canada. But makers are problem solvers! Daniel redesigned his Sawppy to use wooden beams instead.

    Quinn Morley

    Quinn is a person who thinks big. Quinn has been working to scale up Sawppy's design, making modifications as needed, for a much larger rover. This picture shows one of his big wheels next to a standard sized Sawppy wheel. This rover will be a mobile testbed for his project developing deep ice drilling techniques.

    Bob Krause, Inventor Studios

    Bob leads a FIRST Robotics team and they have their own version of Sawppy's CAD file filled with their improvements for a better fit with what they want their rover to do. They can't use their rover in an actual FIRST competition because I didn't design Sawppy around competition rules. However, the team members are thinking about using Sawppy's construction technique for their future projects. All of this makes me extremely happy.

    Chris Dakin

    Sawppy has a sibling in the UK! During Sawppy's evolution I had to adjust various pieces and occasionally aluminum extrusion lengths. This resulted in some recuts and some unnecessary (in hindsight) waste. When I documented their final dimensions, I thought it was possible someone could cut all the required pieces out of two extrusion beams at the 2 meter standard length sold by Misumi. Thanks to the build log for Chris' Sawppy, I saw this hypothesis has been proven.

    Marco Walther

    Marco has lots of robot building experience and wanted to add a Sawppy to his robot army. I received many helpful pieces of feedback about my posted instructions for Sawppy, pointing out where I could...

    Read more »

  • Magnetically Attached Flag Pole

    Roger04/09/2019 at 21:56 0 comments

    One of the problems I didn’t foresee in designing Sawppy was that some children might see a fun challenge in doing running jumps over my rover. I first saw this unwelcome behavior when I brought Sawppy to Long Beach, and I knew it’d only be a matter of time before a child would misjudge their jump and smash Sawppy into a pile of broken rover pieces.

    Clearly I need to find some way to discourage this behavior, but I also can’t do anything that physically harms misbehaving children. This eliminates straightforward solutions such as a Samurai blade pointing straight up. I would also like this countermeasure to be stealthy and not call attention to its anti-jump purpose otherwise some would see it as a challenge.

    The answer came while preparing for this year’s Southern California Linux Expo (SCaLE 17x). I was scheduled to host the Hackaday x Tindie Birds of a Feather meetup, and I was also slated to co-present a talk with Lan Dang. For publicity purposes I decided Sawppy can be a rolling billboard, as SCaLE is exactly the right audience of people who would pay attention to a 3D-printed rover running about. I pulled out a yardstick I had on hand and started planning how to use it as a flag pole, and I immediately knew I had my anti-jumpover countermeasure as well. Two birds, one stone.

    As previously mentioned, I didn’t want this flag pole to be too rigidly attached. If someone bumps my sign, or if someone decides to try jumping over my rover anyway, the flag pole must break away cleanly without damaging the person or the rover. For SCaLE I used a zip tie that was arranged so there is tension holding the yardstick flag pole in place, but pops free when stressed.

    This mostly worked, but as it was built on a balance of opposing forces, it was finicky to reinstall. At Caltech Science for March, a curious toddler yanked off the flag pole and the toddler’s supervising elderly adult tried to reinstall the pole. But grandpa had no idea what he was doing, blindly stabbing inside Sawppy’s equipment bay with the yardstick applying more force as he grew more agitated. After two attempts at saying “Don’t worry, I’ll put it back myself” while I watched in horror at Sawppy electronics getting pummeled, I forcibly grabbed the yardstick from his hands in order to save Sawppy from being stabbed to death.

    Clearly, a better solution must be found.

    In preparation for Sawppy’s appearance at Yuri’s Night, I decided to try a magnetic mounting system. Originally rejected because I thought it wouldn’t be strong enough, I thought it was worth a second look. I had a stack of these powerful little magnets and a single pair wasn’t able to hold the pole. But four pairs of them might be strong enough for the task.

    For this test, four magnets were held on to chassis beam via packing tape. Two on upper beam, two lower.

    Matching sets were held to yard stick flag pole with more tape.

    With updated camera mast and flag pole mount, Sawppy was ready for its next public appearance at Yuri’s Night Los Angeles! That event got crowded at times and there were a few accidental bumps that triggered a clean separation followed by quick re-installation. And while this adult-focused event had few children about, there were plenty of drunken misbehaving adults. The flagpole did not discourage all misguided behavior, but it has worked well enough to become a permanent fixture of my future Sawppy public appearances.

    (Cross-posted to

  • New Camera Mast for Kinect V1

    Roger04/08/2019 at 19:12 0 comments

    When Sawppy first started running around, I wanted something to sit atop a camera mast where the real rover has a camera and sensor array. It is the anthropomorphic head of the rover and it looks slightly wrong without one. (Like a chicken running around without its head.) The first iteration of camera mast sensor array enlisted a standard USB webcam sitting alongside a Google AIY Vision kit. It was mostly for appearance because there wasn’t much software behind it.

    The webcam was fun for entertaining children and occasional longer distance driving, but not immediately useful for autonomy. The AIY Vision box is optimized for classification tasks. I thought there might exist code useful for robotic visual localization but if it’s out there I have yet to find it.

    The most promising tool at hand for rover localization is my Kinect sensor bar running RTAB-Map or some similar software. So Sawppy will inch towards autonomy by getting a camera mast upgrade to my Kinect V1 sensor bar and see if we can integrate that into rover systems in a useful way.

    I went looking for a good way to mount a Kinect bar to Sawppy. I disassembled its base looking for a good mounting mechanism, but there weren’t convenient existing fasteners for me to use and there weren’t good places for me to drill and tap new ones. It was surprisingly crowded in there! I knew there was a motor for up/down tilt but I underestimated size of the motor gearbox inside.

    I then reassembled the base and went with plan B: a simple flat platform for attaching my Kinect sensor bar with double-sided foam tape.

    Kinect sits slightly offset camera mast center for two reasons.

    1. The Kinect sensor bar is very wide and if mounted centrally it overhangs to the right. I worry about it hitting obstacles so I wanted to bring it closer to the middle.
    2. By offsetting sideways, I could expose the top of the pipe used as camera mast and run Kinect’s wire down the middle for cleaner wire management.

    At the moment this Kinect is no more functional for autonomy than the previous configuration… in fact, for its first public appearance at Yuri’s Night 2019 it is not even electrically connected to anything. It’s just a matter of taking one step at a time.

    (Cross-posted to

  • Yuri’s Night LA 2019

    Roger04/08/2019 at 00:35 0 comments

    I'm happy to see interest in Sawppy continue to grow and my little rover now has something of a publicity appearance schedule. Summaries of the past few appearances have been cross-posted on my personal blog as well as this project page, but going forward I think I should keep this project page focused on technical evolution and skip the non-technical stuff.

    As the first implementation of this new policy, here's the summary for Sawppy's attendance at Yuri's Night Los Angeles 2019. Anyone who's interested can follow the link to read more, others can skip this log entry without having to do a lot of scrolling.

View all 40 project logs

  • 1
    New Home for Sawppy Build Instructions

    After a loss of data event for instructions posted to, Sawppy the Rover's build instructions are now hosted on Github.

View all instructions

Enjoy this project?



Marco Walther wrote 08/27/2020 at 05:44 point

If somebody looked at, (a really descriptive name) is the fork of the robot software working with the Sawppy/SGVHAK_Rover web interface;-)

  Are you sure? yes | no

Brett wrote 06/23/2020 at 02:29 point

Any look at adding suspension on this build? One preliminary idea would be to extend the directional rods lightly and place a heavy duty spring round the shaft with a rubber bump to prevent plastic on plastic collisions? 

  Are you sure? yes | no

Roger wrote 06/23/2020 at 03:50 point

Mars rovers are optimized for slow crawling, where shocks are minor and could be absorbed by the curved wheel spokes. They don't just look cool, they act as springs too!

  Are you sure? yes | no

Ameer wrote 06/24/2020 at 03:01 point

Great point roger,

I wonder if Flex wheels would do well (100% infill). or if only tires are Flex and Wheels/Spokes are PLA.

  Are you sure? yes | no

Steve wrote 08/28/2020 at 17:47 point

I've been hacking around on different "Airless" tire designs with this OpenSCAD script here:

This had led me to this intermediate step here:

It's not quite a fully flexible, squashable replacement wheel, but it could be a nice way to "winterize" your rover :)

  Are you sure? yes | no

Ameer wrote 06/20/2020 at 20:38 point


Is there a way to cut thin grooves into the steel wheel shafts and shave flat surfaces in the middle of the said steel shaft without a Lathe Machine that is worth hundreds of dollars?

Is it possible to buy the 10 shafts pre-cut and pre carved somewhere?

I already replaced the Rocker and the Bogie shafts with Screws and Locking Nuts and they work flawlessly. As for the wheel shafts, I don't have a work-around method.

I have been stuck at this stage for a week now. Help!

  Are you sure? yes | no

Roger wrote 06/20/2020 at 21:03 point

Hi Ameer, the shafts are documented on this Sawppy instruction page: ( There's a section on the grooves, with a link to a low-cost lathe alternative ("poor man's lathe") using a drill and a high speed motor ("Dremel") tool. (

I investigated ordering a CNC run of Sawppy shafts and packaging them up for sale, but I would need to order thousands of them to amortize setup cost so it was not cost effective. I'm also watching various Sawppy builders' innovations in this area, lots of improvements on my design. Check out this hex drive by @TeamSG . (

  Are you sure? yes | no

Ameer wrote 06/20/2020 at 21:22 point

Hi, thanks I looked at that one, I didn't think it would be high precision but I guess I will do it that way since it doesn't need much precision I guess. Thanks man

  Are you sure? yes | no

lmckeega wrote 07/12/2020 at 04:38 point

I used an angle grinder with metal cutoff wheel to make the cuts. And a metal grinding wheel to make the flat surfaces. I used a Dremel tool with a the metal cutoff wheel to make the slots for the E clips.

The angle grinder and wheels can be purchased for under $30 at Harbor Freight.

  Are you sure? yes | no

Ameer wrote 07/12/2020 at 20:27 point

I also ended up using the metal grinder with a grinding disc, and a dremel tool.
But I replaced 4 of the shafts with Screws that fit the Barings properly and Locking nuts.

  Are you sure? yes | no

Ed wrote 06/11/2020 at 18:14 point

Hi @Roger . First of all, congrats again for the project. But, above anything, for taking the time to share with the world. I am sure you motivate lots of people to learn more about several areas, mainly because we have something exciting to learn from. I hope to get the point I can contribute as well. :)

By the way, have you ever considered doing a video explaining your solution of the Ackermann steering geometry calculations? That is an essential part of the project and it would be great understanding it a little more. I am doing some research, since me and my kid want to understand the concepts instead of only using or replicating it. But understanding your approach would be cool. I've been trying to get it from the code but it is harder that I thought. 


  Are you sure? yes | no

Roger wrote 06/11/2020 at 18:36 point

Thanks for the kind words, @Ed. It's always rewarding for me to hear from enthusiastic rover builders.

Wikipedia covers general concepts of Ackermann steering ( and the JPL Open Source Rover documentation goes into more detail on how the concept is applied to a six-wheel drive, four-wheel steering rover. ( I couldn't create diagrams with angle and geometry in code comments, so I hope these documents will be helpful for you and your kid.

Sawppy's code is different from the formula in JPL's PDF, but the concepts are the same. I have to admit I need a refresher course occasionally myself. If you are familiar with Jupyter notebooks, I have shared one here (

Finally: don't underestimate the power of good old pencil and paper. (

  Are you sure? yes | no

Ed wrote 06/12/2020 at 03:04 point

:D Thank you very much for the references. 

I am on it right now. 



  Are you sure? yes | no


[this comment has been deleted]

Roger wrote 05/31/2020 at 20:20 point

Ideally the 8mm shaft should be snug and not easy to move sideways within the rod support halves, though once the rover is assembled, there wouldn't be much sideways force upon them anyway. Which means a rover will work properly even if the center differential pivot (rod support halves and friends) is slightly loose.

3D printers are not precision instruments and each one is a little different from another. I tried to design Sawppy to be tolerant of these differences, but inevitably parts would not fit correctly when printed on another printer. Minor modifications may be necessary to adjust for specific printers, which is what @Ameer had to do.

  Are you sure? yes | no

Ed wrote 06/03/2020 at 21:09 point

Thank you @Roger 
And I am sorry I deleted my question above accidentally 😔so I will replicate it here:

Basically, I asked if the 8mm shaft of the Differential should be tight or loose (allowing for lateral movement) in the Rod Support parts. 

Thank you again. 

  Are you sure? yes | no

Ameer wrote 06/02/2020 at 18:15 point

I dealt with this last week:

You can force the 8mm rod into them, just tapper its edge.

The problem is the resulting plastic rod once you join the two parts:

this resulting rod doesn't get stuck inside the barrings and is loose.

The other problem is the two barring's around that piece are also loose because the holes for them are larger than they need to be!

I redesigned all three parts but haven't printed them yet as my printer is busy for another 48 hours.

What's the status on these parts with you?

  Are you sure? yes | no

Ed wrote 06/03/2020 at 21:05 point

Thank you @Ameer !!

Yeah, actually I have printed those parts and I have ordered the rods and they haven't arrived.

But this was a question that came to my mind when checking the how to assemble it.

I will share here when I try to assemble them.

  Are you sure? yes | no

HolgerDK wrote 05/01/2020 at 07:44 point

Hej Roger,

THX for this amazing project !!!

I've a question regarding the body assembly. I can see that the Rocker Body Mount element  and  the Bogie Body element are not equipped with bearings, and as I understand are these the only elements carrying  axes without. Is there any reason for that?

Greetings from  Denmark

  Are you sure? yes | no

Roger wrote 05/03/2020 at 19:51 point

It has been my pleasure to share Sawppy to rover fans all over the world, thank you for the kind words.

To answer your question: the two parts you listed are half of a joint and each have a counterpart with bearings for rotation. So I felt additional bearings would be redundant adding unnecessary cost and complexity. 

Bogie-Body holds 8mm shaft fixed, which fits into bearings of Bogie-Wheels.

Rocker-Body holds 8mm shaft fixed, which fits into bearings of Rocker.

If you disagree with this design decision, please outline your reasoning and we can discuss it as possible future improvement for Sawppy.

  Are you sure? yes | no

HolgerDK wrote 05/03/2020 at 21:20 point

Hej Roger,

THX for your  answer! I'm not a mechanical one. I don't have knowledge to agree or disagree, I was just wondering about this break in the design philosophy. If there are no reports of damage in this part, it will  probably work :-)

  Are you sure? yes | no

Roger wrote 05/07/2020 at 01:54 point

Sounds good! I think it will make more sense once you start assembly. Have fun rover-building.

  Are you sure? yes | no

Diego D. Santiago wrote 04/18/2020 at 00:48 point

Hi Roger, First of all, thanks for sharing this great project! I am wondering if there is any version or remix of sawppy using 2020 profiles and common stepper motors (maybe nema 17). After considering your design, the motors and the extrusion selection are optimal, but in my country this material is really difficult to buy. Therefore, ordinary 3D printer materials may be better for my build.
Thanks again. Great job!

  Are you sure? yes | no

Roger wrote 04/18/2020 at 03:38 point

Hi Diego, thank you for your interest and your kind words. Sorry to hear the components I used for my Sawppy are hard for you to get. I'm not aware of any Sawppy variants using 20mm extrusion profiles or NEMA 17 motors.

20mm x 20mm is stronger but also heavier and more expensive than what Sawppy needed, so I used 15mm x 15mm. I had thought they would be just as easy to purchase, so thank you for letting me know that's not true worldwide. If there are more feedback wanting a different extrusion profile, perhaps they would be motivation for a matching Sawppy variant.

As for NEMA 17 motors, the most popular and affordable types commonly found in 3D printers only have ~10% to ~30% of the holding torque of LX-16A servos. NEMA 17 with similar torque are significantly more expensive and harder to find. Using cheap NEMA 17 motors would require additional complexity, some sort of gear reduction. it is on my personal to-do list but not very high on it. It is likely someone else will attempt it before I get around to it.

Higher on my to-do list is a Sawppy revamp allowing the use of standard commodity RC servos which are far more common than LX-16A serial bus servos. Mechanical changes are minor and a steering prototype was successful ( but the corresponding electrical and software changes have not yet been designed.

Hopefully later this year!

  Are you sure? yes | no

Diego D. Santiago wrote 04/18/2020 at 17:17 point

Hello Roger. Thanks for your answer. How about ordinary aluminum rectangular tubes? they are certainly lighter, will they have the necessary resistance? . I would love to experiment with stepper motors and some kind of reduction if necessary. I know that CAD files are available, however I can't find a way to export them in any format that is easy to modify in INVENTOR. In any case, it may be more convenient to re-model the parts to the correct size. Is there any specification of sawppy geometric parameters? It would help a lot the cause. Thank you!!

  Are you sure? yes | no

Roger wrote 04/19/2020 at 00:32 point

I think metal tubes (either square or round profile) are worthwhile investigation as a replacement for extrusion beams. Fastening to them will require different techniques, and structural strength will depend on wall thickness. And there are probably other problems that will be discovered in the investigation. Please let me know how it goes!

You don't have to use Onshape if you are more comfortable with Autodesk Inventor. If you want to re-model parts from scratch, the overall dimensions are in the "Overlay Sketch" tab of the Onshape document. I started with a diagram from a JPL published paper that had Curiosity side view with major dimensions. ("Reference Image" sheet) I scaled it down a little less than 1:4 (500mm wheels to 120mm wheels) and then measured all the other dimensions relative to that scale. They are in the "Parts layout side" sheet. From there, I extrapolated the top view dimensions and they are in the "Parts layout top" sheet.

You will need to create a free Onshape account to access those dimensional data details. They seem to be inaccessible in the public (no login) shared view. You also need to be logged in to an Onshape account in order to export CAD data to Autodesk.

If you're not opposed to learning Onshape, though, I highly recommend it especially in an educational setting. It opens up full power CAD capabilities to students that can't afford computer meeting hardware requirements of SolidWorks, Inventor, etc. A $200USD Chromebook can be a functional CAD workstation, and I think that's amazing. I chose to design & share Sawppy in Onshape because it is so accessible to everyone.

  Are you sure? yes | no

hernangarage wrote 05/01/2020 at 23:39 point

Diego agree with you... round or square regular aluminium tubes, are going to cost 10% of the design, but a major screw redesign is going to be need. I was thinking on circular tubes, just to emule the original mars Rover boggie frame. In terms of Nema 17, I think is a no-go, since the motors are not prepared to reach speeds near 100rpm which are needed to see sawppy running propertly. I will move to a 3d printed "planetary Gearbox reduction" near 1:16 to use standard cheap HP printed brushed 24v motors.

  Are you sure? yes | no

Roger wrote 05/03/2020 at 20:00 point

I'm happy alternate approaches are under investigation and I look forward to hearing about NEMA17 and metal tube experiments. On a technical note: the LX-16A maximum rotation speed is only 60 RPM. Which makes my Sawppy slower than human walking speed but still tremendously faster than the real Curiosity rover. Curiosity has a maximum speed of approximately 4cm/sec which translates to roughly 1.5 RPM. (People would quickly get bored with a rover moving realistically slow...)

  Are you sure? yes | no

hernangarage wrote 05/01/2020 at 23:40 point

Diego, I'm also from Argentina... I would like to work together... :) look for me on IG. 

  Are you sure? yes | no

RCP1 wrote 04/14/2020 at 22:02 point

Hi Roger,

amazing project! I am currently also planning to build one. But I am a little bit stuck with the aluminum extrusions. I am located in Germany and Misumi delivers in Europe only to corporate customers (which I am not). Do you know any alternative distributor of such extrusions with same dimensions?

  Are you sure? yes | no

Roger wrote 04/14/2020 at 22:57 point

I'm sad to hear Misumi Europe is not as hobbyist friendly as Misumi USA. I have no first hand experience with alternatives but I'm happy to share what information I have.

I've heard from one team building a Sawppy using MakerBeam XL. Supposedly some initial tests are promising, but that rover hasn't been assembled yet. A search for German distributor pointed to this site:

If you build a Sawppy with MakerBeam XL please let me know your experience and I'll add it to Sawppy documentation as reference for future rover builders.

I know of one Sawppy builder in Germany, and I see you've already found Martin and posted a question to that project page. We'll have to wait for Martin to check that page and write a response.

In general Sawppy design should be compatible with any aluminum extrusion beams with a square 15mm x 15mm profile. Most such beams are compatible with generic M3 fasteners, and even if not, hopefully the cost increase and modifications would be minor. 

  Are you sure? yes | no

RCP1 wrote 04/18/2020 at 09:34 point

MakerBeam XL seems to be the right tip! Thanks for that! The German distributor also offers custom cuts, which is awesome! I will try to build one with MakerBeam XL extrusions and share my experience with you.

For others, who are searching for extrusion candidates: Another one seems to be OpenBeam (, if MakerBeam XL and Misumi are not available.

  Are you sure? yes | no

HolgerDK wrote 04/25/2020 at 13:15 point
Hej RCP1,

 I also started now with MakerBeamXL, working  fine. I'm using the M3 nuts MakerBeam provides for them

Greetings from Denmark :-)

  Are you sure? yes | no

Rhys wrote 04/27/2020 at 22:38 point

I can also confirm MakerBeamXL works well. My Sawppy is assembled using MakerBeamXL fixed with M3 square nuts instead of the MakerBeam custom T-slot fittings. I used a fine tooth tenon saw to cut the extrusion (second hand Nobex picture frame saw with a 24 TPI blade).  

  Are you sure? yes | no

Roger wrote 05/03/2020 at 19:38 point

Thanks for the confirmation on MakerBeamXL. I'll put this into Sawppy documentation soon. (UPDATE: Added to

  Are you sure? yes | no

Ameer wrote 04/14/2020 at 19:23 point

  Are you sure? yes | no

Roger wrote 04/14/2020 at 22:42 point

Yay, looking forward to another Sawppy sibling taking form.

  Are you sure? yes | no

Rhys wrote 04/04/2020 at 23:37 point

Roger, I'm in the process of upgrading the controller on my Sawppy from an Arduino Mega 2560 to a Teensy 4.0 and am having an issue getting the serial communication between the Teensy and the LX16A servo board to work. I am using a bi-direction logic level shifter to manage the 3.3V to 5V level shift.

I've established the same setup works for other 5V serial devices such as a  serial RC receiver, just not the servo board.

I have also found that I cannot get a response from the LX16A servo board when connecting to its serial port with a USB-to-UART bridge rather than connecting directly to the servo board's USB port.

I'm wondering whether anyone using this servo board for their Sawppy has seen similar issues?

  Are you sure? yes | no

Roger wrote 04/05/2020 at 02:35 point

The serial communications issues I've known about are classic serial communication stuff. (TX/RX reversal, forgot to connect common ground, etc.) I haven't heard any problems specific to the servo board. However, I haven't heard from anyone who has interfaced its serial pins directly via 3.3V logic so you may be the pioneer here.

  Are you sure? yes | no

Ameer wrote 03/31/2020 at 20:34 point

Roger, while printing I noticed in the list

"4 Steering Knuckle" Are those 2, then mirrored into another 2?


  Are you sure? yes | no

Roger wrote 03/31/2020 at 21:03 point

My own Sawppy has four identical (non-mirrored) knuckles. I designed all functional dimensions to be symmetric so they should work in either original or mirrored orientation, because I expect some rover builders may want to mirror them as a personal aesthetic choice much like the wheels. (Even if the knuckle details are pretty hard to see once the rover is assembled, unlike the wheels.)

  Are you sure? yes | no


[this comment has been deleted]

Roger wrote 03/26/2020 at 00:22 point

This is mentioned in Sawppy wheel printing instructions.

"The rover will need 6 wheels. Printing half of them in a left/right mirror-image is an optional aesthetic choice. Note that the real Curiosity rover does not left/right mirror its wheel spoke design."

Curiosity spokes are the same on all six wheels, see JPL video:

  Are you sure? yes | no

Ameer wrote 03/27/2020 at 21:04 point

awesome thanks. just read it, I deleted the note

  Are you sure? yes | no

Roger wrote 03/27/2020 at 22:07 point

No need to delete. It's a honest question and it received an answer. It's fine to leave it up as reference for future Sawppy builders.

  Are you sure? yes | no

Ameer wrote 03/14/2020 at 20:47 point

From the list of Parts, where do we buy: 2 "RC car steering turnbuckle" that Connect differential bar to rocker joint? Thank you.

  Are you sure? yes | no

Roger wrote 03/14/2020 at 21:21 point

Those are fairly generic items, the important attribute is the end holes must fit a M3 fastener. Here is one example: (affiliate link)

  Are you sure? yes | no

Ameer wrote 03/14/2020 at 22:04 point

Thank you, since we only need 2, are those suitable?
The difference is the min/max Distance eye-to-eye.

  Are you sure? yes | no

Roger wrote 03/15/2020 at 03:27 point

Eye-to-eye distance is not important, since Sawppy's differential bar position can be adjusted front/back to suit. That item will probably work. However I recommend getting extras as spares due to my first hand experience: If not properly tightened, the center portion will work loose and a lock nut will drop off and disappear. And if your luck is as bad as mine, it'll be the reverse-threaded nut.

  Are you sure? yes | no

Ameer wrote 03/01/2020 at 23:01 point

Roger, an FOV Camera streaming the footage to the operator - like any drone

  Are you sure? yes | no

Roger wrote 03/03/2020 at 09:01 point

Self-contained FPV drone camera systems operate independent of the Pi.

  Are you sure? yes | no

Ameer wrote 03/05/2020 at 23:39 point

I agree.
I will attach a Wifi camera on an Alt/Az fork. Capable of Live Streaming and Recording; both in Natural Color and in IR. Controlled with an App on the Same Android Phone controlling the Rover.

  Are you sure? yes | no

Ameer wrote 02/26/2020 at 18:43 point

Hello everyone. I'm building a sawppy or two; I just ordered some parts; whatever that is 'Available' at the moment. I want the robot to do a few things besides move, so I have Questions if anyone can chime in plz.

Best method of a mini HD Camera on Sawppy? Like this size

Can Sawppy handle a small Robotic Arm? Best way to go about it? What if I attach one of these

Thank you in advance

  Are you sure? yes | no

Roger wrote 02/26/2020 at 19:03 point

Camera: lots of flexibility here for a Sawppy rover.

GoPro-style camera mounted on rover will record good footage of rover adventures.

Cameras sold for FPV (first person viewpoint) remote control vehicles are fun for a live rover's-eye-view as you drive Sawppy.

I've used a USB webcam plugged in to Pi on my Sawppy to transmit over rover network, but the Pi wasn't very good at video compression so I had to choose between high latency (bad) or high bandwidth consumption (bad).

I've heard from people who want to do on board machine vision so Sawppy could drive autonomously, and I'm eager to hear results of their experiments.

  Are you sure? yes | no

Ameer wrote 02/26/2020 at 19:47 point

Thank you Roger.
You think a cam should be independent of the Pi?

  Are you sure? yes | no

Roger wrote 02/29/2020 at 16:01 point

It depends on your intent for your camera. Are you recording for later viewing? Do you want to enjoy a live view? Or do you want the rover to process vision itself?

It is also valid to have multiple cameras aboard, catering to different intents.

  Are you sure? yes | no

Roger wrote 02/26/2020 at 19:06 point

Robot arm: Sawppy was designed from the start to have a robot arm mounted in the front like the real rovers do, though I have yet to build one for my own rover. The chassis should be able to handle the type of arms you linked, but you will need to design an adapter to mount their robot arm base to the aluminum extrusion beams of Sawppy's body.

  Are you sure? yes | no

Ameer wrote 02/26/2020 at 19:45 point

Thanks man! Great

  Are you sure? yes | no

lmckeega wrote 03/27/2020 at 19:09 point

My version of the Sawppy, CJ, has an arm. I don't have the skills to program it to be controllable at this time, but it can extend and do pre-programmed movements. Not unlike the real Curiosity, I guess.

The arm uses 5 of the LX16A servos.

I would be willing to share my stl or scad files . 

  Are you sure? yes | no

Ameer wrote 03/27/2020 at 21:05 point

wow. awesome, i'm still wondering how to configure the Arm and the Camera! I'm still printing parts for now, done w the wheels, i'll print more parts this coming week. I should prob. print the servo coupler and other parts you changed, I saw one STL on ur page, are there more? what else should I change in the print?

I'm in Disco Bay, where about are you?

  Are you sure? yes | no

Roger wrote 03/29/2020 at 19:11 point

Have you posted any video of CJ's arm running through some pre-programmed movements? I'd love to see it.

  Are you sure? yes | no

Roger wrote 05/03/2020 at 19:37 point

CJ's arm looks fantastic!

  Are you sure? yes | no

Steve wrote 06/01/2020 at 07:09 point

I'd love to see the CAD for CJ's arm :)
I plan to add an arm to my own Tenacity rover some day:
I even have an old coffee-grinder motor in a box here, waiting to be part of a  home-built RAT.

  Are you sure? yes | no


[this comment has been deleted]

Roger wrote 02/17/2020 at 06:17 point

Thank you for your interest in Sawppy! If you have questions please I encourage you to ask them here publicly. Others can help with public questions if I'm otherwise distracted as I was this past week. If you prefer to communicate privately, you may click on my name and click "Send a Private Message".

I can tell you there hasn't been a "final" version of Sawppy. Things are continuing to evolve under the hands of the Sawppy builders community and myself. If you care to share more information about your deadline, budget, skill level etc. perhaps that can help generate useful suggestions.

  Are you sure? yes | no

Ameer wrote 02/17/2020 at 22:47 point

Thanks Roger,

I will read further on the development and come back with ideas.

  Are you sure? yes | no

Juhwan Yoo wrote 12/27/2019 at 18:17 point

Hi, @Roger Firstly I really appreciate to you. It's so amazing robot. : ) Do you have any plan to upgrade the Sawppy Rover for ROS? In order to control this robot on the ROS environment, I think it's enough to just develop control node with rosserial. (e.g. Jetson Nano installed ROS --> Arduino --> debug board --> LX-16 actuator). How do you thnk about?

  Are you sure? yes | no

Roger wrote 12/28/2019 at 03:51 point

I plan to experiment with ROS on Sawppy using Raspbery Pi --> Debug board --> LX-16 network. I've looked at @Marco Walther ROS project (Thanks, Marco!) and have some ideas on how I can contribute with a different approach. I plan to start with ROS Melodic, which has provisions for Python 3 absent from earlier versions of ROS.

The first planned node will be a Python 3 script subscribed to /cmd_vel and send out commands to all ten LX-16A servos (four steering, six driving.) Once that is successful, I'll build out other pieces of Sawppy functionality.

Can't a Jetson Nano communicate with Debug Board directly? I'm curious why you proposed using an Arduino between them.

Also, NVIDIA's Isaac SDK seems to be a robot framework tailored to take maximum advantage of their products like the Jetson Nano. Are you aware of Isaac? If so, I'm interested in your reasons to use ROS instead of Isaac.

  Are you sure? yes | no

Marco Walther wrote 12/28/2019 at 04:08 point

Yeah, my Rover node is basically a very thin interface to your original Sawppy Python code (plus my handling for the drive motors;-) But all that should easily translate into Python3

  Are you sure? yes | no

Juhwan Yoo wrote 12/28/2019 at 16:31 point

Thanks for your reply @Roger
In fact, there is no special reason to use Arduino. I want to use what you set up code and H/W, without any modifications. In addition, despite of many benefits of Jetson Nano, there are major limitations on the Jetson Nano. For example, it occurs sometimes I/O issues, because of a small size of on board memory. So I want to just reduce some loads on the Jetson Nano. It's all.
Actually I don't know well about ISSAC SDK. Maybe it would be so excellent. But there are not enough materials than ROS to me.

  Are you sure? yes | no

Roger wrote 01/31/2020 at 22:43 point

Hi @Juhwan Yoo you may be interested in this: @Rhys has written a complete software stack for running a Sawppy rover on ROS Melodic. I've just posted a project log here about it. Feel free to read it or you can go straight to the repository:

  Are you sure? yes | no

Juhwan Yoo wrote 12/27/2019 at 18:10 point

Hello, @Marco Walther Could you tell me how to launch and control the Sawppy Rover on the ROS melodic environment? In addition, Can I use the browser to control remotely the Sawppy Rover on ROS? For reference, I'm using the SBC like Jetson Nano, ZED Depth Camera, and LiDAR(LDS-01 for Turtlebot3).

  Are you sure? yes | no

Marco Walther wrote 12/27/2019 at 18:49 point

Hello @Juhwan Yoo ;-) I did not try with Melodic but I don't actually expect many problems. I'm somewhat dependent on my Up-Board. The code is mostly simple Python2. I don't have access to the rover right now, but I'm planning to push all my latest changes to the repo after the New Year.

On the browser control, it should not be too hard to add/modify the browser code to emulate a joystick node in ROS. I did it for my RC setup;-) But I'm not really interested in a browser UI;-) Iff I need that, I can simply switch to the [semi-] original Sawppy Python code and that works just fine;-)

Have fun,

-- Marco

  Are you sure? yes | no

Juhwan Yoo wrote 12/28/2019 at 16:38 point

Thanks for reply @Marco Walther 

Wow, I am very excited the information about your latest code. If possible, for someone like me, please update the Readme content. Actually it's so hard to understand the current code, especially how to use it.

  Are you sure? yes | no

Marco Walther wrote 10/16/2019 at 15:56 point

A little update on my Sawppy. I added the Intel RealSense T265/D435 cameras ( )and I'm playing with ROS[1] on it now. I can drive it via ROS nodes and the rover behaves pretty much as before. I got both cameras to report in the correct frame yesterday. Now I have to get it to map some rooms;-)

  Are you sure? yes | no

Patrick wrote 10/13/2019 at 21:12 point

Swappy, my take on the Sawppy rover design, was a huge hit in the Mini Maker Faire Rocklin! Here's a photo of it at the event:  Unfortunately I was kept too busy answering questions to remember to take any other photos of it, but it was so much fun meeting other people interested in it, hopefully inspiring some to try similar projects on their own!

I did have a few repair sessions to retighten the set screws on the wheels, so I'll definitely be trying your suggestions of loctite and making bigger detents on the axels

  Are you sure? yes | no

Roger wrote 10/13/2019 at 21:35 point

Congratulations on a successful outing! I frequently find myself facing the same problem you encountered: have so much fun talking to people about the rover we forget to take pictures. But as far as problems go, that's a pretty good one to have, right?

I look forward to reading about your upcoming modification experiments.

  Are you sure? yes | no

Marco Walther wrote 10/16/2019 at 15:44 point

I was thinking about the Mini Maker Faire Rocklin;-) But we had fun with a High Altitude Balloon launch & recovery on that Saturday;-)

Too much fun and too little time:-(

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates