Close

Project Log 76: Mechs are not viable... In certain ways.

A project log for DIY Mech/Exoskeleton suit.

Mechs are not viable, nor cheap, so I will try to design and build one alone anyway.

fulanodetailFulanoDetail 01/17/2024 at 00:232 Comments
16/01/2024, Tuesday, 20:47

I'm not dying, by the way, just the project, lol.

Basically, I was recalculating everything over and over again because since day one I had a flea behind my ear.

First I thought "making a mech is actually somewhat easy, they just didn't take this seriously enough because they have more useful things, like a forklift", but being serious, I didn't really think this was a good enough reason for that, but couldn't pin-point why.

Then while searching for artificial muscles and their linear speeds, I found out that the human muscle can achieve linear speeds of over 40 centimeters per second.

And when I put in the numbers, I understood my mistake.


Converting the 40cm/s of linear speed with the 1/10 mechanical disadvantage of the human body in a forearm with 30cm of length, the rpm of the arm would reach 250 RPM, or 4 meters per second of linear speed (I heard that a punch can reach 45 km per hour of linear speed).

Now, if we input these numbers in an mech arm with 60 cm of length (since the entire mech is twice the size of an average human), and we maintain the linear speed of 4 meters per second, which would give 130 rpm, we would need 6000 newton-meters to lift 1000kg.
By the way, it doesn't matter if the load is put in a difference distance, if it is going to lift 1000kg on the tip of 60cm, the lever arm length won't change the final wattage, no matter how you tweak the rpm and torque, you can't scape the laws of physics.

And, if you input the numbers in a torque to horsepower calculator, it gives you 81,686 watts, or 109 horsepower.

If we take the more "whatever" approach and simply say that the arm and shoulder are the same, the torso uses the same amount of energy and the legs use 3 times more, it would still give 545 horsepower, or 408,750 watts.

This is basically half a megawatt.

If we take a somewhat precise approach and simply multiply the other limbs force requirement by 3, the numbers go bonkers.

109 hp to the forearm + (109x3) to the shoulder, ((109x3)x3) to the torso, (((109x3)x3)x3) to the legs, which would give: 

4360 horsepower or 3,270,000 watts of power.

You would need 3 megawatts of power to a simple mech that moves at same speed as a human being.

It is not "super speed", just "super strength", you would still run miserable 10 km/h with the mech (average human running speed), maybe 20km/h due to the double size of the limbs (a mech even slower than this could be viable, but what would be the point?)

And a exosuit with 100kg of lifting capacity would use 145 horsepower/109,000 watts, not much lower than the mech (a exosuit even slower than this could be viable, but what would be the point?²).

Plus, I'm calculating only a 100% efficient machine, which is not possible. With inneficiencies, the energy consumption would increase even more.

In any manner:

My apologies, I thought mechs were viable due to a math mistake I've made.

I will left the project on so anyone enthusiastic enough (like me) will just skip all the homework.

Anyway, it was cool to read about all types of subjects, including artificial muscles and the sorts, and I don't really think I will stop looking at this subject.

This mech/exosuit thingie was more like a hobby were I was having a lot of fun searching about this stuff.

But even so, I feel kinda stupid in a manner. Like I wasted my time due to a silly mistake.

Live and Learn, I suppose.

(by the way, the mechs in Armored Core's 6 lightest Mechs weight around 35 tons)

But, as a consolation prize, maybe walkers will be viable, since they don't need to carry the extra weight the arms and torso would need to carry. xD

Maybe some day I actually try to make a walker-forklift hybrid, but who knows...



The only thing that makes me question whenever or not this math is sound is the human body itself, we consume around 300 watts of energy during activities. Which completely boggles my mind.

I'd say that it is dependant on the type of actuator that you use.

So, since electric motors need 3270 watts to lift 1 kilogram at same speed of the human body, you would need to find an actuator that uses at least 100 times less watts to lift 1 kilogram in order to make a mech viable.

A good competitor is Dielectric Elastomer Actuators, the dielectric pump or the Steam actuator idea, which uses miliwatts to lift kilograms. But, as I am tired of repeting myself, I'm too broke to make and test any kind of actuator.


Oh, by the way², you would spend 35 to 40 horsepower for every 100 kilograms if you were using drone motors to lift your body weight, so you could jump around like you are in zero gravity (or with super human strength, lol).

This would make 260 watts per kilogram, but you would need 4 times more power to to make things move 4 meters a second. Dunno if it would make sense to insert a lot of propellers in a humanoid mech.





I always add off-topic after off-topic on my project logs...

But what I want to say is: thanks for anyone that commented, liked and helped me with this project.

It was fun. :)





Well, one thing I noticed: the Boston Dynamics robot Atlas consumes around 3000 watts-hour during exercises.

Since the robot carries a 3000 watt-hour battery and can only do it for 1 hour, then I'm assuming it uses around 3000 watts per hour. It can move at around 2.5 meters per second and carry maximum 14 kg of load while the robot itself can weight around 90kg to 195kg.

It is hard to find a definitive answer, I can only find scattered information through news articles about the subject. The website from boston dynamics also doesn't give clear information on the subject either.

https://bostondynamics.com/atlas/

On top of that, there are other versions of Atlas, some weight almost 200kg, the shorty white one weights around 90kg.

In either case, I do think it to be strange for this robot to be able to only consume 3000 watts while carrying weight and it might indicate that in fact I'm calculating something wrong.

But taking the 81.686 watts I calculated before and take the force and speed out of it, it makes a little more sense. The robot carries 10 times less weight and at 1.6 less speed, so in total it would be 5105 watts or 6.8 horsepower.

However, the robot is said to only consume 3000 watts or 4 horsepower without the stewart platform skeleton where the load is proportionally divided by all joints.

Of course, there can be a lot of information that is not specified, including the way the robot moves, the way it distribute its loads and so on.
Not to mention that the marketing department simply took approximated values and putted them into the website.
After all, a stepper motor may have a rated speed and torque, but its actual capabilities can only be shown into a graph.

But... I think it is realistic to assume that the estimate I've made is correct.


In the end, it is indeed it is 81.6 watts per kilogram of weight at 4 meters per second of speed.

A simply legged exosuit would consume around 8 kilowatts and 10 horsepower for every 100kg of weight being lifted.

Not great, not bad, I'd rather say.

Discussions

Esteban wrote 01/28/2024 at 22:05 point

Didn't Boston Dynamics do this? Atlas moves at human speed, it probably is not as strong as a human, but it can lift itself up (jump).
Your physics doesn't seem right, it takes about 4.0 watts per kilogram (on average 240 watts) to walk (according to https://jneuroengrehab.biomedcentral.com/articles/10.1186/1743-0003-11-80 , surely the efficiency of muscles cannot be that great compared to hydraulic pistons, especially considering 25% of the energy we consume ends up as kinetic. (For a car, 20% of the energy of gasoline is used) (Granted, the good we eat is VERY energy dense...)
(https://phys.libretexts.org/Bookshelves/Conceptual_Physics/Book%3A_Body_Physics_-_Motion_to_Metabolism_(Davis)/10%3A_Powering_the_Body/10.09%3A_Efficiency_of_the_Human_Body)

Don't give up!

  Are you sure? yes | no

FulanoDetail wrote 01/28/2024 at 22:44 point

For the little I could find about Atlas from Boston Dynamics, and it is said to consume 3000 watt-hour during activities (like running and jumping) and its battery lasts for 2 hours.

If my math were to be right, it would only be capable of lifting 1 kilogram, but it can lift 10 to 14 kilograms, it seems like you are correct.

https://www.wevolver.com/specs/atlas.robot
https://www.techbriefs.com/component/content/article/31078-darpa-80-99s-upgraded-atlas-robot-h

  Are you sure? yes | no