Team 95: Project Semmelweis

We on 95 wanted to do something to help with the Covid-19 crisis like anyone else. However, we did not want to create a solution for a problem that did not exist, or worse: potentially spread infection under the guise of being safe.

So we waited.

We did not have to wait too long though. A local doctor from DHMC got in touch with us through the grape vine. She had an idea for a robot that could reduce risk to their staff and conserve PPE usage and wanted help turning her idea into reality. We were absolutely thrilled at the opportunity. We named the effort “Project Semmelweis” after Ignaz Semmelweis, one of the pioneers in the field of antiseptic practices.

-Deliver and retrieve mobile tablet stands to and from patient rooms without exposing the patient to hospital staff (and vice-versa)

Tablets are commonly used in telehealth to allow patients to talk with their doctors, nurses, and family.

-Robot is easy to sanitize with wipes
-Navigate a potentially cluttered isolation room (IV stands, bedside monitoring equipment, etc.)
-Precisely place tablet stand
-Navigate linoleum flooring with minor defects/obstacles without damage
-Multi-hour run time
-Little or no risk of contaminating the patient room

-Easy-to-learn controls
-Easy to maintain
-Quiet operation
-Drive at walking pace for longer hallway trips
-8-hour run time (one full shift)
-Awesome paint job

… okay, so I made up that last one.

Here are some pictures of the development, without any context.

We have a lot less than our normal level of documentation owing to the fact that most testing was conducted in an active hospital and we were not comfortable taking any pictures or video for the sake of privacy.

The short version of the development story is…

*We brainstormed and sketched out a bunch of different approaches as a team over slack.
*We settled on using a switchable permanent magnet device, in this case an inspection magnetic block, to grapple with a magnetic target we could install on the stand. No moving parts outside of the robot besides the wheels dramatically reduced our chance of tangling with anything.
*Mk1 was a box shape with belt drive using NEO motors and Colson wheels. The drive part of the hardware worked, but was clunky to package. It was what we had on-hand and we learned a lot by building something really fast (we failed early and failed well). We had plenty of time to react.

*Mk2 was powder-coated for easier cleaning and triangular-ish to improve turning clearance in the room. We also put a simple suspension in our magnetic target to handle slight flooring imperfections more robustly. We ran into some assembly and service annoyances with this design. We also identified the velcro-down lid as a decontamination nightmare.

*Mk3 features improved service access, thru-lid charger port, a smooth magnetic lid seal, and a slight tilt in the magnetic target that causes weight to be shifted from the tablet stand to the robot’s drive wheels, improving robot traction and mobility. This is the version we have delivered.


Team members worked as they wanted to through Slack. Manufacturing was donated by one of our big sponsors, Progressive Manufacturing. Powder coating was donated by a local shop called Race Metalsmiths. Assembly was conducted at Grasshopper Machine Werks (my garage/side business). Team alumni primarily handled assembly at GMW, we were not comfortable asking team members/minors to go out and about. @kaszeta, a former 95 coach, handled all of the programming and will be publishing a Git Repo (right?) in a little bit. My day-job, FujiFilm Dimatix, covered basically all of my time on this project too, which was just awesome.

We got parts donations from numerous FRC suppliers: AndyMark, CTRE, and Rev. We used parts from the above and Vex.

Our BOM is here.

OnShape model is here, including a whole lot of scratch work.

More documentation is forthcoming.

As usual for 95 our design is open. Please feel free to copy it, modify it, and use it. Just let us know and credit your source. Please contact me via PM for any special requests.

Any questions and comments are welcome. I know this is a little light on detail for one of my build posts, time is tight these days!


Can you cite in the manual or Q&A where that magnet is legal?


It’s a special type of pneumatic device. It moves metallic things by reducing the air gap between them.


Well Done!


Team 95 continues to be a leader in FIRST. Well done!!


The code isn’t anything fancy (a simple joystick mixer and PWM outputs for the switchable magnets and tank drive), but it’s posted to the Team95 repository:

Note that this is a .Net micro framework program running on a CTRE Hero board.


There were a handful of other people who had positive input on this project that I did not mention earlier (my apologies!)


I am sure that I am missing some… it has been a long effort with poor note-taking. Please PM me if you’re reading this and haven’t been mentioned yet!


Lies! I’ve never had positive input on anything in my life. I won’t have you ruining my reputation. I only ever besmirch and defame according to reputable sources on these forums.


The second I saw 95 and project I knew it was gonna be something spectacular. Amazing work 95!!


This project has good intentions, and likely has some safety benefits in keeping people away from other people. But, the medical industry is a highly regulated one, and equipment that is used in medical facilities is highly regulated and tightly controlled, and for good reason. I would be a bit weary of this inducing a new trip/fall hazard, or a potential hazard with the battery and power system, if something were to ever short circuit. Proceed cautiously. Nonetheless, it’s great that your team is continuing to use this time at home to learn new things and make useful things.


That is a good point that I am glad you brought up.

I am not an insurer or lawyer, I just have some experience with business insurance and liability

In a situation where a company designs, builds, and then markets a device that company shoulders most, if not all, the liability for its use. In a situation where a company builds a part/device/machine to their customer’s specification the customer bears most or all of the liability for its usage assuming that the end product meets the customer’s specifications.

We are operating in the latter condition. This is one of the (main) reasons why we waited to be approached instead of imagining what a helpful robot would look like and pushing that out. This situation allows the project to be covered by the insurance that my side business carries.

I would NOT recommend that any other teams/groups/individuals pursue this kind of effort without careful and due consideration.

Regarding short-circuiting: everything is wired with 10awg, and protected by a 20A main breaker and 30A branch breakers (a bit backwards, yes, but conservative).


In all seriousness we are hoping to use the servo-switched-mag base device on an FRC robot at some point. The mag base costs $15 and is rated for something like 176lbf of pull. One could imagine some fun uses for such a device!


In my experience they’re usually pretty heavy though (easily more than a CIM). Have you found one that’s lighter?

I could not say exactly how much it weighs… they are heavy. It is a big slug of magnet and steel with a little aluminum. We are routinely underweight, more so with the prevalence of brushless motors. So we can probably afford to be a little sloppy with one actuator.

Seems to me there was a really neat ground effect idea recently… Be careful of pulling the retaining pins out of the arena. (Maybe just ask forgiveness afterwards) :wink:

1 Like

Quick update: the hospital has approved the robots through their own internal analyses for deployment to three departments. They are hoping to go live soon.


Quick update (again): the first robot has been deployed to CHaD (Children’s Hospital at Dartmouth). Woo!

We ran into an interesting annoyance before other deployments could be completed. On linoleum floors that see low traffic and/or are waxed regularly the robots drive fine. On worn/unwaxed linoleum they make the worlds loudest nails-on-chalkboard sound I’ve ever heard in my life. The little skid in the rear screeches at an unacceptably high volume. The solution? Mount the ball on a compliant interface instead of directly to the aluminum chassis. We patched things up with a squishy tape to get by and ordered some rubber grommets and spacers to make a nice, permanent, solution.

Here are the three undeployed robots at rest.


Great work on this, as usual for most everything 95 does

As a member of a team that developed a similar device for our local hospitals, I have a few technical questions based on some of the challenges we faced:

  • What is the range on the controllers? How well does it deal with walls between the driver and robot?
  • Is there any feedback for the driver to tell where they are / communicate with patients?
  • How long does it run on one battery? How long does it take to charge/swap batteries?
  • Does it have surge/short protection standard for hospital electronics?
  • Does it have a safety light/beep to warn people?
  • What’s the gross weight? Can it be manually moved if it dies?

They are BT, we were able to drive over 100ft away. Unable to run out of range during any of the trial runs.

During real use the driver will have an iPad in the stand that can be connected to a video call to talk with the patient. All patient rooms have at least a section of glass wall that the driver can easily see through to navigate.

I did 5 hours of testing on one battery and wore it down to around 12.4V… we anticipate 8-12 hours of usage (a whole shift). They recharge with a 1A smart charger, so that will take a little while, but they’ll have essentially all night to do so.

I do not know exactly what the hospital’s standards are for that sort of thing, I do know that it was cleared by their own internal checks though.

Not specifically, though it is loud enough to do that on its own. The usage of these robots will be with the operator walking along side the robot until it is deployed into a patient room.

I never weighed it, but I’d guess <30lbs. Easily moved by most people, and the drive train will free-wheel without much complaint.


I see, the driver stands in the hall while the robot goes into the patient room. Our robot was designed so the operator would stay in a waiting area completely outside the hospital wing where coronavirus patients were quarantined, so the driver wouldn’t have to wear any PPE or face any risk of infection. Obviously that leads to very different engineering requirements.