the last time i spoke with Bill Peduto over the summer. It was an exit interview of sorts, closing out eight years serving as mayor of Pittsburgh. This week, Carnegie Mellon University (CMU) announced that the politician has joined the Heinz College of Information Systems and Public Policy as a Distinguished Executive in Residence. He will serve as a professor and help teach a “mini course” at the university.
The Pittsburgh he leaves behind as mayor is radically different from the depressed Rust Belt town many imagine at the turn of the millennium. It’s a place whose economic transformation into a technology and medical hub has been fueled by a couple of universities (CMU and the University of Pittsburgh) and the UPMC health system.
We kick off this week’s newsletter by catching up with Peduto and discussing how robotics, automation and AI can help shape the cities of the future.
How have these last few months been treating you?
January was kind of a blur. February, I started to put together the S-Corporation. That was throwing money at the tax attorneys and accountants and getting the proposals written and approved. In March, I started doing some of the work on that and then waited to hear back from CMU. It has been a transition. That’s the best way to put it.
We have seen tremendous economic change in Pittsburgh. What role has the fact of having two important world-class universities played in the process?
I think it was Senator Moynihan who said it: If you want to build a world-class city, build a university 100 years ago. It is absolutely the reason that Pittsburgh was able to come back the way it did, between the universities and the hospitals. In the wake of the depression of the 1980s and 1990s, they took the place of institutional heavy industries and were able to become both the economic helm and the engine. They created new industries around innovation that not only took the place and number of people employed in those industries, but also, by creating within those new industries, they were able to put Pittsburgh back on the global stage.
One of the policies he implemented was to allow Uber to pilot self-driving cars on the streets of Pittsburgh. What role have deals like that played in your plans?
2007, 2008 is when I first met and really started working with Red Whittaker. It was before he won the DARPA Challenge, and around the same time, within my municipal district, CMU was using vehicles on city streets in a very defined area. They were also in Hazelwood on tracks. If I hadn’t known, worked and felt that level of comfort, I never would have allowed Uber to hit the streets of Pittsburgh in 2015 and Pittsburgh wouldn’t have been the first city to have self-driving cars on its streets. That would have meant that we would not have seen the several billion dollars in investments and the employment of 5,000 people in building an entire industry here.
Specifically, what role do you think automation and robotics play in the future of cities?
Within Pittsburgh, it is the autonomy of everything. It is not simply about robotics in vehicles, transportation or storage use. It is the autonomous of almost anything you can think of. This is how it works in the future of transversality. How robotics works with medicine, different types of technology, and different emerging fields. What I do believe is that cities like Pittsburgh, which are small and in these emerging fields, have an opportunity to come together and talk and share what they’re doing. I think that’s where we have the ability to see new industries flourish.
Do you think that robotics and automation help to recover national manufacturing?
I see that it is a catalyst not only for automation and robotics, but also for artificial intelligence. I see these companies that haven’t updated for 60 years, or even longer, Callahan on Tommy Boy. Those kinds of businesses are the foundational business of a city and finding the right investors who are willing to say, ‘This business needs to stay, because it supports this community.’ And partnering with universities that are willing to say, ‘we can provide the technology to be able to upgrade’ and partnering with federal government programs that have come through the Biden administration to provide AI, robotics and automation so we can secure another 100 years that the company will be able to continue producing.
Just as TechCrunch is doing its Austin City Spotlight (I wrote a bit about the making here), there’s a lot going on in the skies over the Dallas/Fort Worth metro area. Last week, Flytrex announced it was coming to Granbury, Texas, and now Wing has started drone deliveries in nearby Frisco and Little Elm. The Alphabet company will deliver health products from Walgreens, along with first aid kits from Texas Health and ice cream from Blue Bell, with the promise that the latter won’t melt along the way.
The new addition is Wing’s second US market and fifth overall, joining Christiansburg, Virginia; Helsinki, Finland; and Canberra and Logan, Australia. Availability is limited at launch and is set to expand to more customers in the future. Progress in the world of drone delivery has been somewhat mixed in recent years, though Wing is arguably making the most steady progress, one moderate market at a time.
There’s some interesting recent news worth noting at MIT in the world of accessibility and eldercare robots. A CSAIL team is working on a system that uses a robotic arm to help people get dressed. The problem is one of robotic vision, specifically finding a method to give the system a better view of the human arm it is working to dress.
A new article details the team’s work with a “state estimation algorithm,” which helps the robot predict the position and angle of the human arm during the process. “If the arm is straight, then the robot will follow a straight line; if the arm is bent, the robot will have to curve around the elbow.” Michael Gienger of the Honda Research Institute Europe says in a statement: “If the elbow estimate is wrong, the robot could decide on a move that would create excessive and unsafe force.”
CSAIL also joined forces with MIT, CMU, and UC San Diego to develop “complex mass manipulation,” my new favorite English phrase. The new system, known as DiffSkill, algorithmically teaches a robot how to work with pizza dough in a simulation. Schools add:
It then trains a “student” machine learning model that learns abstract ideas about when and how to execute each skill it needs during the task, such as using a rolling pin. With this knowledge, the system reasons how to execute the skills to complete the entire task.
Meanwhile, Devin wrote about an EPFL team in Switzerland that is working on the very red-haired raspberry picking process. As much as he loves a good strawberry, it’s nice to see some of his other berry siblings get some love from robotics researchers.
“It’s an exciting dilemma for us as robotics engineers,” said Professor Josie Hughes of the project. “Raspberry harvest season is so short and the fruit is so valuable that wasting it is simply not an option. Additionally, the cost and logistical challenges of testing different options in the field are prohibitive. That is why we decided to carry out our tests in the laboratory and develop a raspberry replica to train harvesting robots.”
Speaking of cool robotic arms interacting with famously brittle fruits, here’s a team of Japanese researchers working on a system that can peel bananas. It’s sort of the antithesis of the robot that can put jackets on people, with an even easier subject to bruise. The system uses imitation learning, trained for around 13 hours, with a success rate of around 57%. So there is a long way to go before anyone’s banana peeling jobs are in real jeopardy.
Lastly, ticket sales are open for our Robotics event in July. We’ve been scheduling it for a few months now, and it’s seriously killing me that we can’t tell you about the people we’ve lined up. It’s easily going to be our best yet.
Tickets are always open (and free) for Actuator.