Human Power, Past And Future

We will assume you’ve seen The Matrix — it was from 1999, after all. The surprise, at the end, was that humans were being used as human batteries to power a civilization of intelligent machines. But aside from just putting out some heat, the idea does have some precedent. After all, humans powered machines like mills, sewing machines, and pumps for centuries before there were good alternatives.


Galley ship
Reconstruction of a squadron of ancient Greek galley ships.

Early machines used hand cranks, treadwheels, treadles, and even pedal power to harness energy from humans. Consider, for example, an ancient galley ship with many oarsmen providing an engine. This wasn’t a great use of human power. An oarsman on a galley used his arms and back but didn’t much use his legs. The legs, though, have larger muscles and are often stronger. A pedal boat or racing shell would have been much more efficient, but without mass production of strong metal parts, it would have been difficult to build and maintain such machines in ancient times.

There was a time when pedals or treadles operated lots of machines from sewing machines to lathes. There were even old radios able to transmit and receive with no external power thanks to pedals as late as the 1940s.

Pedal-powered radio
This pedal-powered radio transceiver found use in a lighthouse around 1946.

Today most of what we pedal are bicycles, and most often as a leisure activity. We also use treadmills, but we use them for a different purpose than generating motion through human power. In fact, most treadmills today move using a motor so you can feel like you are running without going anywhere.


That was almost the case back in Victorian England where prisoners sentenced to hard labor were made to run on a treadmill as a form of punishment. In 1818 it was decided that prisoners sentenced to hard labor should have to labor all the time, so they were put on treadmills that did nothing. There were also crank machines that are just what they sound like: a machine with a crank that does nothing.

By 1895 there were 39 treadmills and 29 cranks in use around England. Some prisons eventually adopted the wasted labor to mill grain or pump water, but many were just “grinding the wind” serving no purpose but to punish the inmate.

Treadmill at Pentonville Prison
Prisoners endlessly climb the “stairs” formed by a wheel in Pentonville Prison around 1895.

The United States also toyed with penal treadmills around 1822, but they were never very popular. Typically, these treadmills were set up as wheels configured as endless staircases and had partitions to prevent prisoners from communicating with adjacent prisoners. A fifteen-minute stint on the wheel earned a five-minute break and this went on for up to six hours a day.

Modern Times

Since the rise of electric motors — not to mention changing conditions in prisons — there hasn’t been much interest in using humans to power machines. Pedaling or using a treadmill today is likely to be for exercise or pleasure and not to provide power. What’s worse is that when a modern machine does try to harvest manual energy, it usually does so to generate electricity which is typically not a very efficient thing to do.

Of course, sometimes you really need electricity. For example, a crank flashlight, phone charger, or emergency radio needs electricity. But if you are trying to, say, pump water, you are better off using the energy directly to do the work than generating electricity and then tasking an electric motor to get the job done.

What Tomorrow May Bring

However, we are seeing a trend lately of electronics that use less and less power. Even tiny watch batteries now last nearly as long as their shelf life thanks to devices that have great power economy and improved battery management systems. As devices sip less power, opportunities to power them from the human body increase.

Granted, the nPower PEG seems to have vanished. The Pavegen system that generates electricity from people walking on a special floor doesn’t seem to generate very much power and is mostly used for tracking footsteps more than producing energy.

But harvesting energy from humans could provide energy for micropowered devices, especially wearable or medical devices. Body heat is an obvious candidate, or — borrowing from Pavegen — some type of generator in your shoe. A few experimental medical devices use blood sugar as fuel. For decades, self-winding wristwatches captured your arm’s motion to keep the clockwork running. Maybe a future smartwatch will boost its battery life using the same method.

To make that practical, you need ultra low power electronics. While we know a few tricks, we probably need to get at least another order of magnitude lower to make human-powered wearable devices more than a novelty.

[Banner image: “Pedal power” by KylaBorg, CC BY 2.0]

54 thoughts on “Human Power, Past And Future

  1. Would be nice, along with solar, wind, if people could get say a treadmill that would also store its energy somewhere. Also when I was young lots of things were manual, like can openers, tools and other common items.

    back then, many people would walk to a local business to get food and cloths and other things. These days, in many places in the US, walking can be dangerous activity with how roads are not designed. Not to mention all these small shops are long gone.

    1. “a treadmill that would also store its energy somewhere.”

      I’m picturing this. It has a giant rubber band inside that twists as you go. It gets harder and harder to keep twisting until finally the user can’t even hold it back any longer. The band un-twists powering the treadmill in reverse. The user is flung backwards across the room.

      I like it!

    2. Once I found a bike with drive and ups to store energy. Problem is that human can’t produce much energy. Few times I was training on orbitrek trying to keep above 120W. It was not easy for more than 15 min. Coffe grindong or Charging tablet/cell phone – fine. Laptop – maybe. Toster – forget.

      Check toster challenge on on youtube :)

      1. It’s worth pointing out that few appliances are as needlessly inefficient as a toaster.
        1500 watts through a resistance coil, OPEN at the top to allow all heat to instantly escape?

        We’ve fixed light bulbs, and most appliances are Energy Star… and then we have toasters.

  2. The greater problem for human power is that food production is extremely inefficient to the point that you will use more energy (and fossil fuels) in total riding a bicycle than a moped.

    Back in antiquity you had about 20:1 energy return from growing food with manual labor, whereas today we’re getting less energy out than we put in at a rate of 1:3. Turns out industrialized agriculture, forced irrigation and synthetic fertilizers didn’t make agriculture more efficient – they just made more food at greater costs. This is known as the marginal cost of production. Making one more of anything eventually overtakes and negates economies of scale.

    1. Yes,
      Walking burns 300 W at 5 km/hr, or 216 kJ/km. The marginal cost of raw food energy is about $0.10/MJ ($1 / 2400 kcal, about a day of food). It costs $0.022 in additional raw calories you burn to walk 1 km. (incidentally producing 0.1 kg CO2). Cycling is better by about a factor of 2.

      An electric car burns about 0.2 kWh / km. Here, last month, that also cost $0.022 / km (though I could carry three friends and some groceries in the car too, for the same cost).

      Gasoline? Well, that comes in around $0.16 / km these days, though half that is taxes not levied on those other road ‘fuels’.

      1. Walking burns 216 kJ/km but your metabolic and mechanical efficiency is around 25% so your real food energy consumption is closer to a kilowatt-hour per km, plus recovery.

        1. No, it’s 300 W food energy input power, the majority of which just goes into warming you up, and much of that gets blown off in evaporating water in respiration or in sweat. As you say, little of it makes it out as useful mechanical work.


            Figure 4.10.2-1 Aerobic endurance. 40% VO2 max estimated food calorie input based on oxygen consumption for an average male comes in at 433 Watts (1480 Btu/Hr)

            >In cycling, for example, the human has an efficiency of about 21%.

            Which puts the output power for a whole working day of cycling at 40% VO2max to 91 Watts. For a hard working person, going up to 50% VO2max could yield 734 Watts input and 154 Watts output for about four hours. At the 60 minute endurance mark, your input power can be 1158 Watts and output 243 Watts – which is btw. why the motor power limit of electric bicycles in the EU is set at 250 Watts – that’s “human power” scale.

            Walking and cycling have approximately the same power demands, though cycling is faster. In any case, if you see someone quote walking at 300 Watts for the total expenditure, they’re more strolling than walking because that would correspond to an oxygen intake level below the lowest point in the quoted figure above.

          2. 40% VO2 max, 433 watts is 12500 kJ (3000 kcal) burned in a single 8-hour shift, a pretty heavy workload.

            Yes, walking at 5 km/hr is not a very high workload.

    2. Really depends on how you define it all – the moped is burning energy and so are people, but it is almost all energy sourced via various degrees of directness from the sun – and the moped has many many more stages of conversion from the suns rays hitting the planet to motive power than a human on any diet, and a vegetarian diet less stages still, every stage introduces losses…

      And of course food production efficiencies are about the most wildly varied thing imaginable, and can be done better than in antiquity if you work the right way, and no doubt even worse than your modern ratio…

      But on the whole a human powered system is better (at least on a individual scale) as it can be considered as a smaller closed cycle. And the extra costs of keeping active humans alive vs sedentary ones is not all that great – infact its probably going to work out lower as the truly sedentary human then needs more other aid to stay living, add in the massive infrastructure needed to maintain and fuel a vehicle (of any sort) which largely isn’t a problem for the mostly self repairing human (or horse I suppose) it becomes more clearly in favour of human motive power in general..

      Its only really the mass transit of people or goods by other means that make human power look both impractical and probably less efficient overall (not counting how much of a hurry everyone is to get places these days, which again rather rules out man powered propulsion most of the time).

      1. Those many stages are irrelevant from the human perspective, because in the end it’s oil in the ground, and we count our EROEI at the point of extracting and refining it – and this is the point where both motor fuel production and industrial agriculture gets its input energy that WE expend. Whatever energy the plants gain from the sun is rather irrelevant, because we’re still spending about three times as much into it.

        You can say food production for human power adds more stages to the fossil fuel inefficiency chain, rather than remove from it.

        1. Not to mention that food itself is inefficient. You could in theory obtain energy from just rice or potatoes, corn – whatever is the cheapest to produce – but that’s not how people eat. You’d become ill. We grow animals to have protein, and processing plant matter into meat substitutes is also highly inefficient (and ends up with a lot of surplus carbs which are fed to Americans).

        2. As I said it depends on how you look at it – Oil’s creation cost is not at all irrelevant from a human perspective anymore either – as we as a species have cottoned on to the fact continuing to burn stored sunlight from millennia ago isn’t good for our longer term viability as a living planet… That means you will soon have to consider all the steps required to fake the oil products from the short term surface carbon cycle, which takes energy, lots of it, including no doubt a great deal of extra energy spent growing more crops for fuel.

          You could indeed consider food production as adding another user for fossil fuel, but its not really another stage as its not built upon stage after stage from anything else fossil fuel powered – its more another parallel branch from the point of extraction/refining, which means no extra inefficiencies from the losses at each stage, just more consumption.

          However the extra cost of producing food for more active people should work out at zero for a very long time – there is so much food waste that wouldn’t end up being waste if people needed to consume more calories so it was eaten before it spoiled…

          1. Nobody’s going to “create oil” – that would be silly. We don’t have to care about how efficiently oil was made because it’s already there. That’s just distracting from the point.

            > its more another parallel branch

            Tractors and harvesters run on diesel fuel, which is derivative of oil, the Haber-Bosch process runs on natural gas, etc. You should consider that we can run a car directly on LNG/CNG – or – we can turn it into ammonia fertilizers, run farms and equipment, grow food, transport the food, refrigerate the food, cook the food, eat the food, and then release what energy is left over by pedaling a bike. Both end with the same result, but using food as the energy carrier has more steps and wastes more of the original supply of gas (and diesel, oil, etc.)

            And the extra food supply problem is a logistics problem, not a supply/demand problem. If people buy more, stores order more to keep the shelves full and this simply scales up the waste problem with it.

          2. Dude there is a growing demand for bio fuels – you really are creating the oil products for those – which takes energy, and is the way the world is going (and really has to go to a large extent if humanity wants to have much of planet to live on and to keep using fossil fuel technologies).

            Also food supply wastage isn’t just a people buy more stores order more scaling up the waste, as a large amount of that waste is in peoples homes for one thing, the food they bought and didn’t eat, and the stores don’t buy excess to waste, they sell it as it gets close to its date cheaply first – which if folks are eating more they are bound to pick up and actually eat..

          3. Also just because you can turn oil/gas into fertiliser doesn’t mean you should… There are many ways to get fertiliser, and ways to vastly reduce the need for it via proper crop rotation for instance…

            How you farm doesn’t have to be stuck in the 70’s when oil seemed like a free, limitless bounty there was no reason not to burn as fast as you can get for everything…

            And my point about it being a parallel – just another consumer is that you were making the crap anyway directly from your crude oil/gas source, its not another processes on another processes just to let you farm – there are no further stacking of process inefficiencies it just increases the demand. Not disputing that your way of considering it has some merits, just making the point that HOW you look at it matters.

            (Also while you can run a car on x the distribution and storage of the huge volumes of whatever fuel you run them on adds into the efficiencies, or lack there of of that method, where an increase in physical activity doesn’t so directly change the transport, storage cooking costs of the food, and the increase is actually pretty small – if you commute to work by bike you would have to be commuting a very long way for the extra energy demand to make a noticeable difference to your food intake – empty that sack of potatoes maybe a whole meal earlier, but it still lasted weeks (and you probably didn’t have to deal with wasted ones)… Plus with the need to be active for your health if you were to instead drive to work and then a gym… Doing the good for your health and commute at the same time is an obvious winner.)

  3. In the original script for The Matrix, the plot was that humans were being used as computer processors as part of a neural network. The studio didn’t think people would be able to grasp the concept and were forced to change it.

    1. That makes so much more sense than the movie.

      Also, we’ve now gone full circle. My grandfather always sent down the (coal) mines as a kid, and the dystopian future sees humans built into (bitcoin / etc) mines.

    2. OMG!
      Although I really like the movie, the ridiculous idea to use humans as an energy source for a hyper-intelligent AI was bothering me to no end. The original script describes it exactly how I thought it would make sense.

      Thanks for restoring my utmost respect for the screenwriters!

      Stupid studio dumbing down such an amazing screenplay -.-

  4. It seems likely to me that we will create synthetic biological devices that will act as an extension of ourselves. This means they will be little neural networks that behave like an organ, being fueled by blood sugar. At some point I figure that a simple biological patch on your brain will provide you with the ability to simply always know what time it is as well as have perfect awareness of balance, among other things. The trick is figuring out how to communicate with it, which I imagine will be something that can be done with guided learning and a fairly simple headset.

  5. I once for fun postulated a gym where the machines generated power.
    The bicycle (also known as spinning bike) is likely the best candidate for electrical generation, but even these wouldn’t be that great.

    A typical person seems to be able to generate 150-300 watts “continuously”, for a couple of hours. (based on a fairly small sample size.)

    But in a gym, a spinning bike is typically not used for hours on end. And most of the time they stand collecting dust.
    But lets say we have a gym where these sees a duty cycle of about 30%, and with an average output of 200 watts. Then for a gym open 24/7 will only generate 1.4 kWh per day per cycle. (though, one can only put in X amount of cycles before there isn’t enough users to use them.)

    Though, with modern LED lights, the spinning bikes could likely keep the lights on in the gym.

    In my own opinion, these are the most logical piece of gym equipment to put a generator onto, not that it would pay for itself anytime soon. At 30 cents per kWh and 1.4 kWh per day we would only generate 153 dollars a year per bike. So it might take a couple of years to pay off the initial investment in generators.

    Then there is the technical side of “just adding a generator” to a bike and have that energy put into the grid.
    Here I would personally go with a small brushless motor (500 watt ones can be had for 100-150$) a bridge rectifier, a cap or two followed by a boost converter to output a more or less “constant current” output onto some 40-48 V power line to make it easy to just parallel them all up. And then carry that DC over to a centralized inverter, with likely a smaller battery bank for “smoothing” out the supply.

    So the bike outputs current, the DC voltage rises on our bus, the inverter/battery-management-system notices the increase in voltage and starts loading it down to keep it regulated, allowing for arbitrary amounts of parallelism for both the generators and inverters. (though, for safety reasons we would likely also have the controller for the boost converter be powered by the external 40-48 volt bus, so that we can have a female connector for the bus/cable and a “exposed” male one on the bike without it being a shock hazard.)

    But in the end, I suspect the economic viability for electrical generation from gym equipment is more or less dubious. The spinning cycle is the best candidate, but even then it likely wouldn’t make sense in a lot of regions.

    1. Back when I worked at Intel, the stationary bikes had USB ports to charge smartphones and tablets. That was the old “plain” USB-A that could supply 10W. With USB-C able to go as high as 240W (although 60W or 100W is more common) with the capability to dynamically adjust the power, it could work well as a quick charger in public areas.

        1. The spec of what a USB type A port on a random device should be able to provide is indeed 500 mA at 5 volts.

          Though, the connector can take 2-2.5 amps without much issue. And some phones will happily try to pull such if the power supply can supply it without major voltage drop. A lot of phones starts limiting the current at about 4.8 to 4.9 volts.

          Together with abusing the voltage tolerance of USB specifications, we can get a fair bit of power over it.
          Since USB allows for 5 volts +/- 5%, ie 5.25 Volts is “fine”, so it isn’t uncommon to see chargers supplying 5.2 volts, and typically up to 2-2.4 amps. ie, 10 to 12.5 watts.

          And that is before any quick charging protocols have been introduced.

    2. I know this is one of those famously flawed ideas for people who haven’t understood the relative scales involved in power generation, but I’m still a fan of seeing all that work do *something* more than generate waste heat. And even a little something is better than that. Maybe I charge a big storage battery (slowly). But the readout can tell me how much energy I’ve stored – in terms of units I can relate to. Like pedal until you’ve charged this AAA cell. Or until you’ve stored the equivalent of eating 1 banana. Or added 10% to my phone battery, or to raise 1 liter of water to 1m elevation and so on. Boiling 1 cup of water becomes a group challenge and watt hours become literal muscle memory.

      1. “every watt is sacred”

        Wow, saving energy is a big waste of energy! Why are you attacking the tiny wastes while ignoring the big ones? Adding insulation to your house is about 10000 times more effective use of your time. Save some real energy.

      2. Every little bit does something.

        But the fun thing with grid scale generation is that the grid is made up of many smaller generators.
        And in terms of loads on the grid, these are typically even smaller and more numerous.

        Collecting power from simple sources like an exercise bike wouldn’t do much on the grid scale. For every million people, not many tends to be on an exercise bike at any given moment, but some are. And even if that just leads to a few kW of power, it still does something.

        I can see a gym noticeably offsetting its power bill through this, but beyond that it won’t really power a city. It is potentially sufficiently economically viable for it to pay for itself and not just be a gimmick, at least for a gym where the equipment is going to see more or less daily use for more than just an hour or two, unlike in a home setting where it might be used 2 hours twice a week.

        I have heard of people needing to queue for the spinning bikes at some gyms during the more popular hours. And the more popular hours is seemingly after people leave work, where the electricity prices also tends to be highest in most cities. So at these times the gym might even be net positive. (Worst case the power might just be sucked up by increasing demand for AC to keep the place comfortable.)

    3. Black mirror episode: Fifteen million merits.
      Everybody’s cycling on a stationary bike, for “merits” (money) needed to live. That’s the closest we can get to your idea :)

      1. If the generator is sufficiently cheap, around 200-300$ and the centralized inverter isn’t too expensive either. Then I could see it as realistic for a gym to have to reduce their power bill if they operate in an area where electricity prices are sufficiently expensive.

        Since a gym would likely have the equipment for a number of years. So even if it only brings in 150 dollars per bike on average, over 5 years that is still 750 dollars each. (and who knows, the added mental bonus of generating useful power might lure people over to use the bikes a bit more.)

        But as a form of power generation, solar panels and batteries are a more economically viable combination in most areas of the world.

        1. Better insulation and a better HVAC system is a far more effective way to save energy. Pet projects are fun if you have unlimited funds but the rest of us care about return on the dollar.

    1. You are comparing apples to oranges. Laptops are usually sleeping at zero power while your machine will probably be on all the time. What about the power for your big monitor, your drives, etc. these things use more power than the CPU.

  6. Giving the fact that a human produces 6 liters of methane daily, this makes a total wasted power of 2.75 Wh (1kg of methane is 2511 kJ, and methane volumic mass is 0.657 kg/m3). Enough to power your smartphone, if you can catch this energy ^^

      1. They also got the amount of energy per kg wrong.
        It is about 55.7 MJ/kg of released energy when burned, though most of this would end up as heat. But to only get 2.5 MJ of “useful” energy out of 55.7 MJ of initial energy is an exceptionally low efficiency.

        Also, talking about gas in terms of volume is a bit inept. (even though the pressure is always considered to be 1 atmosphere for volumetric measurements.) It is far better to talk in terms of weight.

        But I do suspect the typical human produces more than 0.4 grams of methane per day. (22 kJ/day)

        Though, a lot of methane can be extracted from anaerobically breaking down sewage, and if one adds that number onto what gas is produced inside the human body, then our number will be substantially larger than 0.4 grams a day. (though, most sewage treatment plants do not capture methane, and actively mix air into the sewage to greatly reduce anaerobic digestion.)

  7. ” The surprise, at the end, was that humans were being used as human batteries”

    In the end? A surprise? It wasn’t like they saw dead people or anything. That reveal was pretty early on in the movie wasn’t it?

    1. When you see “designed in California” that means they didn’t waste any time worrying about cold climates. MIT got burned by this when they asked a California architect to design a building, the odd roofline captures the snow until it all falls at once right over the entrance.

  8. Hmm, now I’m wondering what would be the best way to human-power a big wooden ship? Doesn’t seem like an easy problem. Also, made much more complicated when you assume war ships, since you want redundancy.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.