MachinePix Weekly #1: Ian Rust, founding engineer at Cruise
Interviewing Ian Rust, founding engineer of Cruise. Also, beetles and controversial pottery.
This issue is the first to feature original long-form content on machines and engineering: an Interview with Ian Rust on building Cruise.
The regular Week in Review following the interview provides behind-the-scenes details on the machines featured this past week. Unfortunately, there is no way to email gifs effectively so I’m using screen-caps (please let me know if you have a solution).
I am still experimenting with the correct way to structure this newsletter, so don’t hesitate to reply with questions, comments, and cool machines.
Interview with Ian Rust
Ian was the founding engineer of Cruise, an autonomous vehicle company acquired by GM in 2016. I have been fortunate to consider him a colleague since he joined Root Ventures as an Entrepreneur-in-Residence.
I sat down with Ian to learn from his reflections on building an autonomous vehicle company and the practice of mechanical engineering.
You started your career in aerospace but quickly switched to robotics before autonomous vehicles. What was driving those decisions?
The main reasons were actually macroeconomic factors. The first two internship seasons I had were right in the midst of the financial crisis [in 2008]. I tend to avoid defense industry stuff, which is where a lot of aerospace falls, due to personal values.
The first year, I was unable to get an internship at all so I just worked a minimum wage job. By the next summer I had started to become more interested in robotics. Coming from the Mechanical Engineering department, I felt that I was lacking experience in software engineering, a critical part of the robotics stack. I was willing to take any engineering internship, and thankfully I was able to find one at Boeing in software engineering.
Six years ago you joined as the founding engineer at Cruise, arguably one of the biggest successes in autonomous vehicles. How did the team come together?
Originally the idea was Kyle Vogt’s, and it was for an aftermarket autopilot for cars. He had started the company and was in the middle of Y-Combinator. The prototype was not complete with 30 days til Demo Day. That’s where I came in: I had known Kyle from MIT and we were part of the same living group and he reached out to me because of my interest in robotics.
I was super-hyped for this idea, so much so that I left Google X before my vesting cliff. I had always been interested in autonomous vehicles - I remember reading about the beginning of the Google autonomous vehicle program [now Waymo] in Popular Science in high school so it was an easy sell to me. I spent those 30 days days working 16 hour days building the first highway-ready prototype.
Cruise has built out a large engineering org and I see their cars zipping by in San Francisco all the time now, but I imagine it was very different in those early days. What were those first few weeks like?
One story sticks out: we were 3D-printing basically all of our hardware that we couldn’t get off of McMaster to prep for Demo Day. It turns out if you leave a car out in the sun in a Mountain View parking lot, you can actually melt PLA. Our printed hardware melted on the prototype and I had a really panicked 24 hours right before meeting investors where I had to rebuild everything.
The challenge of building an engineering organization is threading the needle between when “anything that works as soon as possible” is good to when “we need processes and experts so we aren’t building a house of cards”. It’s different for every project, and being over-extended in either direction is dangerous. I was so happy when we had investors and we were finally able to hire people with expertise besides my own.
It feels like the pace of change in engineering is faster than ever. What advice do you have for mechanical engineering students today?
Practical machine design and mechanical design is very very difficult to teach. Making mistakes and figuring out how to fix them with constraints is core to engineering and hard to implement in classes. You have to have side projects - this is the only way to become an exceptional engineer. If you are a student you have to be exploring outside the curriculum.
I really don’t think you can effectively learn to design and build things in a classroom setting. You need the room to fail and break things. Iteration is good - churning through seemingly stupid ideas is my favorite way to learn and every once in a while, something stops looking stupid.
What about advice for engineers in their careers?
I made two big choices that I think have worked out well: I left Amazon Lab126 for Google X because I had been the only automation engineer on the [Lab126] team. I think it’s important to surround yourself with people that are much better than you. When you’re the best, it’s confidence inspiring at first, but you quickly feel the stagnation.
With Google X, I tried really hard to learn how to work with and around people. I saw that the biggest gains were made by people that could effectively bring in systems integrators and work well with them. Almost all engineering endeavors worth doing require teams, and effectively working with people and creating good team dynamics is often more important than the individual engineering work. Team dysfunction will kill projects faster than anything.
Shameless plug, I’m excited that you joined us at Root Ventures as an Entrepreneur (and Engineer!) in Residence - any particular areas of interest or that you’re exploring as a next side project or career opportunity?
Just for fun I was recently trying to print color on frosting on donuts which is a surprisingly hard challenge. I am also exploring parametric animation because I love video games and animation. There’s still a kernel of artistry required to produce really good things, and I think technology can help that kernel grow faster. I started a calf exoskeleton project that I ended up shelving - I actually got to the point where I was about to put it on my leg but found a design flaw that could have seriously hurt me.
At a higher level, I’m really excited about sensing. A lot of research has been done on computer vision because of the prevalence of cameras. Other sensing modes haven’t been given as much love, but because of autonomous vehicles, there are a ton of sensor technologies being rapidly developed. For example, radar gives you first, second, third order modes of response where you can pull out a lot of nuance. Ultrasonics, 3D, etc - we’re still really early in seeing what we can really do with different sensors and that’s exciting for me.
I’ve also started to explore manufacturing at the edge. We used to have hand tools in centralized production facilities, and we’ve developed a lot of automation there. I think we’ll see an analogous change in smaller decentralized processes like construction and some kinds of food production.
The Week in Review
July 19, 2020: Los Angeles-class submarine USS Hartford (SSN-768) surfacing through the ice in the Arctic Circle.
This was part of ICEX 2018, a biennial British-US joint naval exercise for operating in arctic conditions. A surreal photograph shows the USS Connecticut surfacing not far from the USS Hartford. Follower @williamscraigm asks an interesting question: did the submarines have the chainsaws on board or did the surface crew from Ice Camp Skate bring it them? 🤔
Manufacturer: General Dynamics Electric Boat
July 18, 2020: Stretch hood packaging machine by MSK Tensiontech and Bonus: vibratory testing of stretch hood packaged pallet.
This was a very satisfying find because I’ve always wondered how the pallets are wrapped when I’m shopping at a Costco. The vibratory test was too visually amusing to not include. Errata: MSK Tensiontech is the product, MSK Covertech Group is the company.
Manufacturer: MSK Covertech Group
July 17, 2020: A wireless streaming camera carried by a darkling beetle by the University of Washington.
There’s a surprising amount of history of trying to instrument or control insects. University of Tokyo and Texas A&M tried to remote controlled a cockroach in 1997 and 2015 respectively. To my knowledge, this is the first time someone’s added a powered camera!
Manufacturer: Autonomous Insect Robotics Lab
July 16, 2020: Jizhou ware ceramics from Jiangxi, China.
This was surprisingly the most controversial (and popular) post of the past week. It is floating online misidentified as Raku ware, which is similar but different.
Many people thought the video was a trick and I started to doubt the veracity as well (the cut between the kiln and the glazing shows difference pieces). I was going to delete it until Follower @PUKraus surfaced a 2014 article about French potter Yvon Le Douget who spent twenty years reverse engineering the process and showing you could in fact imprint a leaf on ceramic if you could get it to burn without igniting.
July 15, 2020: Robot for wrapping fiber optics around power lines by Facebook Connectivity.
This post sparked a lot of debate. Why not bury the lines? It’s very expensive. Isn’t this damaging to delicate fiber optics? Follower @marcsavy explains: [fiber optics are] fragile in tension, so this looks like they are coiling into a spiral-like shape to take advantage of bend insensitive fibre's tolerance of bend -- this would provide the ability to 'stretch'.
Manufacturer: Facebook Connectivity
Bonus Time Capsule
July 20, 2015: The MV City of St. Petersburg, a car carrier with a unique shape designed to reduce fuel consumption.
A roll-on/roll-off, or “roro" ship built for Nissan. This odd ship was optimized for aerodynamic drag, saving 800 tons of fuel annually. The scale of ships are pretty crazy: the largest 15 release more nitrogen and sulfur pollution than every car in the world combined.
Manufacturer: Kyokuyo Shipyard Corporation
If you enjoyed this newsletter, forward it to friends (or interesting enemies). I am always looking to connect with interesting people and learn about interesting machines—reach out!
If you know how to non-destructively remove stuck pins from a Wilton vise, definitely reach out.