Whenever I see ISS dockings I wonder why the arm isn't computer controlled. There's no specific information on this page but I'm assuming it was a manual docking. Can anyone explain why the process is still manual?
The cynical part of me thinks because manned spaceflight, at least in the context of being stuck in a LEO tin can, is of questionable utility and we try to find things for people to do? We can automate quite a bit of this, in fact, the ESA's Automated Transfer Vehicle just docks itself. Wikipedia:
At a distance of 249 m, the ATV computers use videometer and telegoniometer data for final approach and docking manoeuvres. The actual docking to Zvezda is fully automatic. If there are any last-minute problems, a pre-programmed sequence of anti-collision manoeuvres, fully independent of the main navigation system, can be activated by the flight engineers aboard the station.
Or what's that apocryphal story about early astronauts fighting with NASA engineers to put in pilot controls with manual override so these guys could claim to be pilots and not mere passengers? The ugly truth is a lot of this stuff is best left to automation, robots, etc.
In the Shuttle, the landing gear control was intentionally left outside computer control so that the vehicle could not be flown unmanned. Flight control software could autoland the shuttle from orbit, but it took a human being to let the wheels down. So I don't think your cynicism is excessive.
Except automation in spaceflight was a solved problem long ago. The Buran had more automation than the STS because of US politics in regard to the STS than some major innovation. In fact, the Buran is the poster-boy of lack of innovation. Its just a slightly souped up COPY of the STS long after the STS was in production. This like calling a 2014 Kia that looks like a 2003 Mercedes innovative because it has the option for a bluetooth stereo.
The Air Force's shuttle is actually autonomous not just automated. Its, say, more like a drone than a R/C car. Its many steps ahead of what the US and the USSR were doing in the 80s. From what we know its a very interesting technical achievement.Over 600 days in space for a spacecraft is pretty impressive also.
>OTV-3 had spent 600 days in orbit by 8 August 2014
I don't know why this metric isn't mentioned more often. This this is practically launch and forget, give occasional orders to, and land once in a while sci-fi spacecraft. From an intel point-of-view this is pretty much a spy satellite that can go anywhere in orbit, change its mission at any time, and then land and go back at any time. It may also have anti-satellite weapons. I guess we won't know its full capabilities until its declassified, but from what we know its a very interesting project from a technical achievement point of view.
I was only being sort of half serious but this thing is amazing, I remember seing amature satelite spotted _photograph_ this thing in orbit on its top secret mission. Glad to know more about it.
I think that was meant mostly as a compliment. Both trying everything and doing the right thing are rare enough behaviors, even by themselves. We should strive to be as good as we were, when he said that.
From an algorithmic point of view, you could rephrase it as "You can always count on Americans to pick the optimal solution by brute-forcing through the entire search space".
Towards the end of the program, a "jumper cable" was carried to get around that:
The cable would connect an avionics bay in Discovery's middeck with the
controls one level up on its flight deck, effectively allowing flight
controllers in Houston to perform landing activities currently done by
shuttle astronauts.
Those manual activities include starting the shuttle's auxiliary power units,
deploying an air data probe, unstowing the orbiter's landing gear and
releasing its drag chute after landing, Herring said.
Best described as a scaling problem with manual or prototype, vs automated.
Lets say you need a prototype ability tool and die maker machinist on staff to handle "things" during development. And you need a simple piece of threaded rod. You could blow a lot of extra time and money on getting a CNC programmer and the software and a numerically controlled lathe or machining center dropped in to make that boring simple little piece of threaded rod. Or you could say, "dude, I know this is beneath your skill, but you're just sitting there burning oxygen and it'll only take ten minutes for you to machine a piece of threaded rod, so ..."
If you have a tool that's designed to do anything, and the tool and op are just sitting there, even if you could automate a one-off, the overall system cost and productivity is higher if the op just does it by hand.
If you have a VERY active schedule, and maybe 3 simultaneously operating 24x7 arms all over the station with only one dude available to run all the arms and everything at 150% of designed thruput capacity etc, then it would make economic sense to automate this task so the arm op can work on something more human oriented, but ...
I bet the trainers that certify an astronaut is precise enough to get it done are a lot cheaper than the engineers required to certify a computer wouldn't karate chop the station.
The arm almost certainly has a control system that can resolve that. Since humans suck at higher order control, the arm almost certainly has a control system that takes the operator's desired output (which will be a series of joint angles), and converts into the required motor torques. In this process, the control system almost certainly also does sanity checks to prevent self-intersections, collision with the space station, overtorqeuing the motors, and etc.
Replacing the human with a computer program therefore likely doesn't literally involve verifying that it will not karate chop the station - lower level control will take care of that for you.
However, as anyone who has worked with robot manipulators before, actually getting good performance out of them in variable environments (ie, not a controlled factory assembly line) is quite a bit of work - work that probably isn't worth it.
considering the first flight where they were rewriting code while it sat outside the station; reconfigured the LIDAR on the fly; I would put it down being overly cautious, maybe a bit of fear? Let us be honest, with all the difficulty of getting that module into space docking isn't that big of a deal. If there were no people on board no one would care.
Plus I have to ask, if a thruster suddenly misfires or goes haywire who is going to be faster? In Apollo days I would put odds on the astronauts, today - I am betting on a computer. We will sooner trust a car to drive itself before we allow a freighter to dock with a space station.
In fact the arm can be remotely controlled. As of 2005[1] ground controllers can uplink sequences of commands for the arm to complete. According to an article[2] about this Dragon launch, DEXTRE (two-armed human-like remote manipulator) will be remotely commanded to unpack cargo from the Dragon's unpressurized trunk and mount it on the station. I believe this is not a new thing at all.
