Cutting hair LIVE (it's hard) (Patreon)

About a month ago I was asked if I would be interested in the following things:

1. Give someone a robot haircut (YES)2. Give the haircut to someone famous (YES!)3. Do it on a livestream with millions of views and massive stakes (uhhhh ok YES!!)

I'm a total noob when it comes to youtube and livestreaming.  Aside from small patreon streams to ~10 people I've never done a livestream before. It wasn't until a week out that I really appreciated how challenging this was going to be...

The thing about going live is that you get one shot, no do-overs, it works or it doesn't. Unfortunately for me, brand new, highly complex, prototype hair cutting robots aren't really the gold standard of reliability. This began to really sink in as the event drew near and the pressure mounted to get the robot finished and working. I also had the challenge of transporting the robot across the country (NC -> LA) with no shop access in the event something broke.

In hindsight this was pretty foolish to attempt, but then again I'm not usually one to let the possibility of epic failure stop me. Spoiler: the robot worked, but the haircut was pretty awful since the host's hair was full of hair gel. The gel made it difficult for the robot to do its thing. But that's OK with me - I'm super happy it didn't crap out which could have easily happened. I'll be doing more haircuts for the video about the project to really show what the machine can do. 

This thing was a beast to transport... (uber drivers weren't super thrilled, but no XLs were available...)

The robot getting there and working wasn't a total accident. Here's how I optimized my odds of success:

1. Pre-baked toolpathsEvery movement that the robot would perform was computed ahead of time and tested over and over with dry runs (no cutting). This removed run-time computed movements from the equation which are a huge source of variability and failure due to the software complexity. Some movements related to probing the head could not be known, but they were bounded and well defined. 

2. Spares of EVERYTHING:This was very annoying and a bit expensive but ended up saving my butt. With a few exceptions I had a spare for every printed part, circuit, trimmers, etc. I ended up needing a couple printed parts, the spare trimmers, and a few wiring components. They are tiny parts but could have been total game enders without them. 

3. Mobile toolkit:Growing up I learned to accomplish everything with a dremel and hand tools. This tool kit took me back to those basics. I had an entire suitcase stuffed to the gills with hand tools, cordless soldering iron (saw a lot more use than I would have liked), other EE tools, dremel, etc. 

4. An obnoxiously expensive caseI just sucked it up and got a huge pelican case for the robot. This was critical because the robot is quite dainty  and I couldn't imagine any other way of packing it up where it wouldn't have substantial risk of injury or death. The pelican case worked beautifully with the robot in perfect condition upon arrival. I'm certain this will see a lot of use in the future...

5. ChecklistsEverything that I did at game time for the livestream was thought out, practiced, and codified in a check list. Things like what order do I turn things on, initializing the microcontroller, and a long list of functionalities to verify. This was good because it ensured I didn't make a dumb mistake and that I adhered to the final point:

6. The final point: NOTHING NEWAnyone who partakes in competitive sports knows that you don't try anything new on race day. New shoes are a recipe for messed up knees on a marathon and new food is a surefire way to have gastric distress. I really didn't want gastric distress so everything that me and the robot would do was tested and verified in the day(s) leading up to the event. 

Checking and re-checking everything

Despite all that prep it almost was a failure. I discovered two major show stopping issues hours before going live. Thanks to my spares and toolkit I was able to patch them and the show could go on.   

This was for the St. Jude charity livestream put on by the Game Theorists. It was very interesting to be part of such an involved production. All the prep / crew / infrastructure is the polar opposite of what I'm used to - just me alone in the basement with a camera. It was a very successful fundraiser and ultimately smashed their goal of $1M, raising $3M for a really great cause. Wowza. 

This concludes an incredibly tight sprint through November. Going into the month I thought it would be no problem to make two videos and finish this robot for the livestream. I told myself that all I had to do was hit some stuff with my bat, make an unpickable lock and a new hair cutting robot. How bad could it be... I've learned that whenever I say that, watch out.  I wasn't wrong about being able to do it but it was very tight. Although I lost a lot of sleep and built up some stress, I can now cross doing a big live stream off my bucket list. I'm also proud of the two videos and looking forward to challenging the lock picking lawyer with an updated version of my lock :)

Thanks everyone for the support, it would be difficult to get these things done without your help. I really appreciate it!