Artemis: Form follows for function robots on the Moon

December 16, 2018

 

 

Recommendation: Read it! The science is as good as it was in The Martian, so enjoy life on the Moon with robots as they show off well-designed morphology and variations of autonomy. 

 

Robots: non-anthropomorphic ground robots

 

 

With his second book, Artemis, Andy Weir has established himself as the new Arthur C. Clarke, writing hard science fiction in a pleasant, digestible way. The serious reviews (versus the “OMG Andy Weir has another book so it must be so good!” ones)  make it sound like the book is about a female Han Solo type in a lunar city. Which doesn’t sound that interesting (why smugglers on the Moon?) or science-y (why smugglers on the Moon?). This poses the real fear that Artemis is a sophomore slump, the dreaded second novel that is a mistake or rushed to market. I know I certainly resisted reading Artemis because I didn’t want to be disappointed.

 

No worries! The function of the book is to explore all the aspects of a lunar city and how it would really work. The form of the book is that it uses a lovable scamp of an engineering genius, who got off track with the wrong boyfriend and is now a smuggler, as the excuse to take us through the city’s entire infrastructure as well as the political, corporate, and social organization- with a few observations about planetary economics and welding in a vacuum thrown in. 

 

So the plot is really secondary, which is a hallmark of hard scifi. If you read Clarke’s A Fall of Moondust, you probably remember all the cool technology and fun facts about rescuing a vessel that has sunk into a moon crater— and none of the characters. I vaguely remember the astronomer who lived onboard a satellite so he could concentrate on his telescope, but not so much as a character rather as a “hmmm, I guess you might send up a person to live by themselves” aside. In hard scifi, the characters and plot serves to get you through the hard science, rather than the science is there to be a realistic background for the nuanced character development and original plot. Is Jazz as likable and relatable as Mark Whatley in The Martian? No, but she’s entertaining enough and her poor life choices don’t distract from the science of what she’s doing. 

 

And speaking of science, there are robots in Artemis! Yay! Robots with realistic morphology and autonomous capabilities! Yay! Indeed, the suspenseful action centers around blowing up robot mining equipment with the help of a small multi-limbed robot. 

 

Like the book, the highly realistic robots follow the "form follows function” principle- and it does this in two ways. The first way is literal: the morphology (shape) of the robots depends on their function. The mining robots are basically large bulldozers, pretty much what terrestrial excavators for pit and strip mining look like now. The hull inspection bot (HIB), which Jazz calls Hibby (of course), is one of my favorite robots of all time. Its form follows its function of performing inspection and maintenance tasks outside the domes, minimizing the number of times people have to suit up and go EVA. Instead of a complex humanoid robot, Hibby is small with limbs that can act like claws to let it easily climb up the rungs that cross the domes, put it has a small arm for manipulating the environment- see my article for Science Robotics on why real space robots like R2D2 need arms here. But it can also move on the lunar soil and inside the dome, adapting its movements to these environments. 

 

The second way the robots follow the "form follows function” principle is in the control style. The robots aren’t some form of Artificial General Intelligence (AGI) but instead their intelligence— as well as the way a person controls and interacts with the robot-- depends on its function.The mining robots are built to autonomously harvest ore; they are not very bright, about the level of a sheep or a roomba, but don’t require supervision for normal operations. A mining robot will plod along doing its job, but if detects a problem, it alerts a human who can then teleoperate the robot in order to identify what the problem is, diagnose the cause and come up with a way to recover, also known in engineering as fault detection, identification, and recovery or FDIR. In the last century of engineering, it was straightforward to identify the problem and the recovery was directly related, similar to a stimulus-response behavior: if green, then go; if X is broken, do Y. But now systems are much more complex and robots interact with the world, which can be the cause of the problem (like an off-road driver suddenly getting high centered on a rock), Therefore, even though the system is “autonomous”, there are times when the system needs to be easy for a person to step in and temporarily take over. 

 

As an aside, autonomous control tends to be optimistic, leading to what cognitive scientist David Woods dubbed (Robin) Murphy’s Law of Autonomy- that an autonomous system will perform at lower levels of autonomy in the field than planned for; a variation on Murphy’s Law that whatever can go wrong will. In my research in field robotics, I emphasize that a robot is never 100% autonomous in the sense of being on its own, because there is always some interaction with a human to specify the mission, report results to AND because robots always encounter a problem that requires a human as a higher cognitive power to sort out.

 

While the mining robots in Artemis illustrated how even an “autonomous" function involves humans, Hibby showed how a “teleoperated” function involves autonomy. No human in their right mind wants to have to tell Hibby how to climb the inspection rungs studding a habitat dome; a human just wants to direct the robot to the desired dome and side of the hull and let the robot do the rest on its own while. Would cowboys really prefer to tell a horse “left side hooves up, right side hooves down” or let the horse do that part. In general, humans prefer— and it is more efficient-- to give high level commands for routine operations such as “return home,” not teleoperate the robot. In theory, Hibby can be considered a teleoperated robot because it is being given directives by an operator, though those directives are often delegating a skill or action sequence to the robot. It is more like Hibby has intermittent autonomy, which is often called semi-autonomy. Jazz and Hibby sometimes work together, with Jazz looking around through Hibby’s sensors while Hibby continues to climb (shared control flavor of semi-autonomy), and they sometimes take turns where Jazz delegates to Hibby, Hibby performs the task and then returns control to the human (traded control flavor of semi-autonomy).

 

Autonomous, semi-autonomous, shmonomous— the distinction between autonomy and semi-autonomy isn’t so clear in field robotics. The goal is for the robot to work and there will be a human involved in that somehow, either to specify the work or to help out. Any robot whose form does not permit human control will not be able to reliably perform its function.

 

Don’t let the description of Artemis hold you back from reading it— its function is hard science fiction, with the form being a slightly implausible plot. Sit back and enjoy life on the Moon with robots. And my recommendation is to get the audiobook- Rosario Dawson is excellent and applies her Batgirl in the Lego Batman movie voice over chops to Artemis.

 

Read Artemis today by ordering with Amazon, just follow the link below:

 

- Robin

 

 

 

Share on Facebook
Share on Twitter
Please reload

  • Grey Twitter Icon
  • Grey Facebook Icon
  • Grey YouTube Icon
  • Grey Instagram Icon
  • Grey Pinterest Icon