Artificial Intelligence
“What to Expect When You’re Expecting Robots: The Future of Human-Robot Collaboration”, by Laura Major and Julie Shah, has some good information, but it suffers from the problem many books by academics have, it is too focused on the narrow specialties of the authors. They almost completely ignore manufacturing robotics, and there are only vague reference to legislation. While this is problematic, there are enough good things in it that it’s worth a read.
The author notes on the back flyleaf show that Laura Major is CTO of a major autonomous driving project, and that she previously worked on drones. The same notes point to Julie Shah’s background in aeronautics and her position at MIT. There is an old adage that you should write about what you know, and the two do just that. The challenge is that they should have spread their wings a bit more. Robotics is richer than they seem to know and collaborative interactions with humans isn’t a concept that is new to vehicles on the road.
Manufacturing, Robotics, and Human Collaboration
The first indication of the problem is Figure 3 on page twenty. While the chart talks about how early automation was worked on in airplanes, their line for “other robotics” (as opposed to aircraft and cars) begins with Roomba in 2002. What kind of computing education fails to note that robotics entered the manufacturing floor in the 1950s? While there’s some great information about what they are learning in the vehicle space, it doesn’t exist in a void, and the authors should have taken a broader view.
What they do cover is interesting. The best parts deal with human-automation interfaces in airplanes. One of the key points they make is that much of the “human error” blamed for aircraft collision and crashes shouldn’t be laid at the feet of the pilots, but at the feet of designers. Chapter 3, “When Robots are too Good”, has some excellent examples of poorly thought out user experience (UX) issues that led to accidents. Other examples pepper the book.
When looking at both pure robotics and collaboration, manufacturing should have been mentioned much earlier and in more depth. For instance, it wasn’t until page 127 that I saw a passing reference to “industrial systems” and page 147 before “industrial applications” was used in a sentence, and neither mentioned robotics.
Manufacturing’s history of robotics allows other sectors to learn from that experience. Robots began as large, very separated, machines on linear assembly lines. As they advanced, machines moved closer together. Collaboration between robots began to be investigated, so spatial awareness was needed. In recent years, human-robot collaboration has been a serious issue, with vendors of such product showing up at major ERP vendor conferences I’ve attended (remember lots of people in a room?).
It isn’t until page 191 that Amazon’s warehouse robotics is mentioned. That’s far too late, as they aren’t the first company looking at that issue, as robots moving off of the assembly line and into warehouses is also something that wasn’t developed just by a few newer companies.
There is much that current companies are learning, but also much knowledge they’ve leveraged from the decades long history in robotics. To think it started with a vacuum cleaner is to do a disservice to the readership.
Legislative Issue For Robotics
The other major issue I had was in the area of legislation. A number of things they presented, they did without thought to legal and social issues, and there one only one place where regulation was really mentioned, towards the end when they described San Francisco limiting the number of delivery robots in each zone of the city.
A much more appropriate place to mention it would have be very early, when they provided an image of lots of robots clogging up sidewalks. Why should they be allowed on sidewalks? Those are for pedestrians. Even bicycles are, in most places, legally restricted to roadways. That’s an issue municipalities will have to address from a legal, regulatory standpoint.
More importantly, “The Reasonable Robot”, reviewed last year, focused solely on legal issue with robotics. Given its dense coverage of autonomous vehicles, I would have expected at least a bit of a discussion by the authors about this important issue. One of the most interesting questions from that book is if it will be a tipping point for the adoption of autonomous vehicles when their accident rate is lower than that of human drivers, so court decisions of fault are based on what a reasonable robot will do.
There was far too little discussion of regulations is a book about society and robotics.
Human-Robot Collaboration, Where the Book Shines
For all my complaints above, where the book does shine is in some of the higher level issues of collaboration. If you understand the authors are focused almost exclusively on roads and homes, what they say is very interesting.
One of the best quotes from the book is “Interacting with an intelligent robot is fundamentally different from interacting with an interface that supports automation.” Understanding that difference will be a key to social acceptance of robotics. Automation is, well, automatic. It is doing exactly what we, whether programmers or users, tell it to do. Robots have their own intent, their own goals. That adds a complexity to interaction.
While concepts about collaboration are throughout the book, “This City is a Cyborg,” chapter 8, has that as the main focus. While it does discuss the possibility of separate lanes on freeways for robots, as mentioned above, it doesn’t discuss any legal reasons why. Still, it’s a high level discussion of some of some key social issues.
The closest to discussing legislation comes in the final chapter, “It takes a village to raise a robot.” The discussion in this chapter is about a clear way that robotic collaboration with humans differs in the world outside of manufacturing. While there are multiple robotic vendors, in a manufacturing floor you almost exclusively see a single vendor. Its robots can communicate with each other in its own protocol, to organize together and to collaborate the same way with humans.
In homes, on roads, and elsewhere, there will be multiple vendors. Each group of robots will have different goals and different communications, making interaction difficult. The authors then bring in the context of aviation. The FAA created an accident database which all industry partners use to input data and perform analysis. Similar and, most likely, far more extensive legislation will be needed to require better collaboration between robotics and humans in the social environment, and that will probably include required collaboration between companies in the sector. It is in this chapter that the book shines and where it has the most value.
This review is longer than most of my articles because the book is frustrating. Human-robot collaboration is a critical issue that needs to be addressed. There are some very annoying things in the book and key pieces missing from it. Still, there’s enough of value to make it worthwhile reading the book. I’ve attempted to present both the good and the bad. Your mileage might vary.
