How was the Hardiman exoskeleton controlled?

[SirFlamenco]

GE’s 1969 exoskeleton was able to lift a thousand pounds, but how did the human control the robot, since the weights involved were much more than the person could lift unassisted? I’m asking this since all modern exoskeletons seem to improve on your strength, while Hardiman did all of the work for you with very little weight transferred to the wearer.

 

[MacGyverS2000]

Overall it was a huge failure, and I don't believe they ever turned it on with a humanin the suit. the main control appeared to be hydraulic (25hp @ 3000psi), with some minor electric (15V @ 3A), likely for hydraulic actuator control.

Hand control was hydromechanical rate control with force feedback. Arm was electrohydraulic bilateral servo control with a 25:1 force feedback ratio. Legs were electrohydraulic unilateral servo with indirect force feedback. Of course, the force feedback was so poor the entire thing would shudder when they tried to use it.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[SirFlamenco]

Thank you for your comment. I already knew that it was controlled by eletrohydraulics. I intend to do build a scaled down version with hydraulic actuators, but instead use DC directional control valves. These would control the pressure of the fluid, and they would be themselves controlled by load cells at key joints over my body. Batteries and subsequently the actuators powered by them currently do not have enough power, speed nor strength to be viable. I think electric valves with hydraulic actuators are a nice compromise, and its implementation would be much easier compared to the 1960s.

Question still remains though, how was the Hardiman controlled? How did the exoskeleton know when the pilot was moving his arm, and especially know when to amplify his strength? The answer might sound simple, but if it's not using pressure sensors, how does it know when to react. Also, if they had to put force feedback to avoid catastrophic results where the user would apply too much force, then it tells me that under normal conditions no force would be put on the pilot.

 

[MacGyverS2000]

This is directly from GE's website (bolding mine):

The complexity of the 30-joint exoskeletal system meant researchers had to develop entirely new control and feedback systems. They apparently accomplished this but their reports don't explain how.

Despite the engineering breakthroughs of the project, the exoskeleton itself never made it past the prototype stage because of its weight, lack of stability and power supply issues. Some unexplained bugs, such as what a 1971 report on the project described as “violent and uncontrollable motion by the machine” when both legs were activated, meant that the project was far from ready for rollout when it ended in August 1971. Hardiman's walking speed was 2.5 feet per second, or just under two miles per hour.

From the pictures, one can make some assumptions, but probably the best you're going to get is their patent:

https://patentimages.storage.googleapis.com/94/b7/51/9496b0adf4aa09/US3535711.pdf

From my quick reading of it, it appears the entire feedback system was hydraulic in nature, with a portion of the hydraulic pressure on the suit sensors being used to provide resistance back to the inner-frame sensors. An acceptable solution (or attempt at one) in the 60's, but hardly the way to go some 50+ years later.

Here are some of Mosher's papers:

http://cyberneticzoo.com/wp-content/uploads/2010/04/GE-Hardiman-paper.pdf

http://cyberneticzoo.com/wp-content/uploads/2010/04/Mosher-Handyman-to-Hardiman-1967.pdf

Dan - Owner

http://www.Hi-TecDesigns.com

 

[MacGyverS2000]

As an amusing sidenote, here's a paper model of the Hardiman suit to print/cut out and assemble

https://static.tumblr.com/ychvj87/sZDn4lec6/hardiman-1.pdf

Dan - Owner

http://www.Hi-TecDesigns.com

 

[SirFlamenco]

What about a more modern exoskeleton such as XOS 2? The best image of the inner workings I could find is this one. After having done a bit more ressearch they appear to be using "force sensors" too. If those aren't load cells, then it must be something quite close, so at least I'm in the good direction with my project. There is a bigger concern though, and it's the mention of a cable system. Normally that wouldn't worry me, but I have found some similar exoskeletons to also use such a solution, and I'm wondering whether or not there is something I'm missing. When I ran some numbers of the speed of hydraulic cylinders, I have found that you need quite a powerful pump to move them at only a low speed, and in my case it was 2 inches per second. This is way too slow for normal usage, and it is why I am building the first version with penumatics. Cheaper and faster, but harder to control and obviously less powerful. Could this cable solution help with the speed and if yes, how would it even achieve such a goal?

 

[MacGyverS2000]

Those are questions only you can answer as only you know the problems you're trying to solve and the methods chosen to solve them.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[SirFlamenco]

Must have not worded it correctly. What I'm wondering is just how cables improve exoskeletons.

 

[MacGyverS2000]

Cables provide a motive method where the driving "engine" is somewhere other than the motive point. In other words, the large/bulky pneumatic cylinder doing the pulling can be situated on the suit's back where it's out of the way rather than right on the wrist (where it adds weight, may get in the way, etc.).

Dan - Owner

http://www.Hi-TecDesigns.com

 

[SirFlamenco]

Ok, so after having done a little bit of ressearch I have compiled a 99% complete list of exoskeletons/bipedal robots using hydraulics from 2008 onwards. Only Nadia, Atlas and Handle are still being worked on. Interesting fact, none of these robots/exos can lift over 200lbs, even though the linear actuators used technically can. So what I think is happening here, is that the joints are the main limitation for the lifting power. They will often use electric motors or some sort of cable solutions hooked up to an hydraulic cylinder. I mean, the HULC legs exoskeletons were technically able to lift 900lbs if they were only limited by the linear actuators, but in practice could barely do 200lbs. So right now I have trying to design an exoskeleton that wouldn't limit your freedom too much while still only using linear actuators. I'm at 20 cylinders right now, but by far the hardest part is how the exo is gonna walk correctly. After all,the exoskeletons currently built haven't had this as a problem since the user can use his own body weight to balance it. In my scenario, we are talking about a complete package of 800lbs+ so it is much more like a manually controlled robot then anything else. I have to look at how robots balance themeselves in ressearch papers, which involves a painful ammount of math. I believe 1 cylinder per ankles could do it, coupled with the two cylinders in the back for a perpetually "falling forward" motion.

 

[MacGyverS2000]

Sitting in one place, you don't need much force to hold the thing in place... gravity tends to do that for you. You only need to overcome gravity (and a bit of friction) to move a piece up/forward. I don't recommend going down the path of self-balancing... if you're running into these types of questions on a near-mechanical, near-static only device, I fear you won't have the capabilities to handle the electronics/math/programming involved to handle a dynamic system. there's a reason why PhDs are made on such subjects.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[SirFlamenco]

Standing still isn't a problem. Effectively distributing the weight while walking is, and compared to other exoskeletons, my body mass is almost irrelevant in the process so I have to see how robots do it, but none of them are made for heavy lifting, so they can use electrical motors and cables without it being an issue. I'm going into uncharted territory here, but that's also what makes this project so exciting. If you're wondering what's the point of using hydraulics if they are so limitied by joints, frame and pumping power well that is the reason for why we only have 2 bipedal hydraulic robots still being worked on, all the rest swtiched to electric after 2014 or so. According to IHMC, the lab currently building Nadia, they can get much more power in a tiny package, which can allow for more degrees of freedom in a small robot. Obviously a non-issue for me though.

 

[MacGyverS2000]

The only thing I wonder at this point is what is your final intention? You seem to be trying to recreate the success of others but aren't acknowledging the downsides to each approach.

My advice is to start building what you think is the right path... you'l soon realize it's not as easy as you believe it is.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[SirFlamenco]

I think you might be a little confused. I'm not trying to recreate anything, as I said this will be the first hydraulic heavy load carrying exoskeleton since Hardiman. The main aspect from other hydraulic robots I can take inspiration from is the walking mechanism.