This isn’t meant to be a treatise on the topic, but an article yesterday on Citizens In Space about the benefits of Human-Tended experiments got me thinking about a chain of thought I’ve been meaning to write about for many years.
I actually agree with the conclusion that in many cases human-tending of experiments is very valuable, especially when it is combined with the easy ability to quickly and affordably refly an experiment. Even in cases where easy and affordable reflight isn’t available, such as the ISS, I’ve heard a lot of anecdotal evidence that many experiments are saved by a little bit of astronaut fiddling/tweaking. However, human-tending combined with quick and affordable reflight opportunities is even better, because such a scenario can allow you to rapidly debug a system that doesn’t have to be designed to be perfect from the start. Existing sounding-rocket flights are so expensive, rare, and reflight opportunities spaced so far apart, that you’re pushed to a situation where the experiment better work the first time or you’re (for instance) going to be changing your dissertation topic, or downgrading to a Masters degree instead of the PhD you intended. When stuff has to be perfect the first time in a challenging environment, it gets really expensive (which shouldn’t be surprising–perfection is never cheap). When iterations become cheap and quickly available, it’s much easier to design something that’s 80-90% of the way there, and tweak it a few times to get it all the way right. So, you can count me as a big fan of what Virgin, XCOR, and Blue Origin are trying to do by opening up the opportunity for human-tended suborbital research.
That said, as most of you know I was one of the early founders of Masten Space Systems which is at least currently focused on unmanned suborbital research flights, and I think that there’s a lot to be said for unmanned suborbital research flights (that have quick and affordable reflight capabilities) as well. First off, an unmanned suborbital research flight can be much more optimized for research than a vehicle that is also piloted and has an experiment tender. Not to mention that an unmanned vehicle is likely smaller, cheaper to replace, and may thus be able to try out riskier experiments than you would want to fly with a pilot on-board (say micro-g cryo-fuel management or transfer experiments for instance). Second, the very ability to manipulate a human-tended experiment in real-time means that you’re also perturbing the experiment. Humans breath, they have pulses, they bump things around even when trying to be careful. The reality is that a human-tended suborbital experiment will have a harder time getting precision pointing or precision microgravity than an untended experiment. Basically while robots are still very crappy at being humans, humans are even worse at pretending to be robots. This is actually a concern that’s already been voiced about the ISS and which is why there’s frequent talk about making some sort of man-tended free-flyer. People are super-useful, but they (and the life support equipment needed to keep them alive) fundamentally screw up the microgravity environment you’re trying to achieve. In many cases you may be getting closer to milli-gravity than microgravity! Third, once an experiment has gotten highly repeatable, there’s a point where humans introduce more experimental error than the benefit they provide by on-site troubleshooting. Robots are great for repetitive processes once they’ve been debugged. Fourth, once flight rates have gotten to the point where pricing starts becoming more competitive, my guess is that robotic suborbital vehicles will often be cheaper to fly than manned ones with human-tending capabilities, if due to nothing other than their much smaller size per given experimental payload.
So in my opinion the reality is that human-tended and robotic suborbital and orbital experiment platforms (especially ones that allow you to do quick and affordable iterations!) are actually highly complimentary technologies. If you only had one or the other, you’d be much worse off than if you had both. The human-tending allows you to quickly debug flight experiments, and verify that you have a repeatable and automatable experiment. The robotic counterparts allow you to switch over to highly-repetitive, cheaper, and higher-quality microgravity and pointing capabilities. Each has its strengths and weaknesses, and I think that as suborbital vehicles like Lynx, SS2, Xogdor, and New Shepard get into operations, you’ll see a natural progression of some experiments starting out on the human-tended side for debugging, but eventually switching to the robotic side for large numbers of repetitive follow-on experiments.
Diversity is a good thing in this case.