Does a bike exist such that you can program it with what load you desire and it will shift gears for you based on that load or maybe never exceed that load? For example, suppose an older person has some knee trouble but doesn't like to shift gears and doesn't want a 1 speed bike so they desire a bike that can be programmed with a maximum load level so that if either the terrain or headwind or hill or any combination would put the bike at a greater load, the bike would automatically sense this and adjust the gear ratio accordingly. For example, someone is on flat paved ground at a comfortable output (perhaps 100 watt equivalent). It would be slick if when that person encounters a slight upgrade that the bike would sense a reduction in speed and calculate the loss of power due to less cadence and adjust so the load feels the same to the rider.

I am also wondering if this algorithm would differ much from constant cadence or if they would be identical or virtually identical. I think it would be significantly different since suppose someone programs in constant cadence of 60 RPMs. With maximum load instead dialed in, if the terrain makes it easy to pedal (and for example there is a tailwind), the rider might be able to use a high gear and pedal slower than 60 RPMs but still maintain desired roadspeed (10 MPH for example). This has happened to me in real life scenario. I think constant cadence would be somewhat annoying since sometimes the circumstances dictate much less load so why pedal at 60 RPMs when in that situation you can get by with 30 or so? It would be nice if the bike would also sense cadence and determine if you "underpedal", it is cuz you are wanting to slow down or cuz the load is light (tailwind for example) so a taller gear can be chosen. Not an easy algorithm to get right in all cases.

  • 1
    A frame with flex, combined with a bottom bracket cable guide filled with dirt, may autoshift on heavy load. Also referred to as ghost shifting. This is not a feature.
    – Criggie
    Commented Apr 10, 2016 at 2:06
  • I assume that if someone pedals with the same amount of effort, the bike should be able to detect this and shift the gears accordingly to maintain what feels like a constant load to the rider. In some cases this will not be possible such as going up a very steep hill or going across some terrain with very high rolling resistance... so the programming of this would have to handle those as well.
    – David
    Commented Apr 10, 2016 at 5:23
  • 1
    @David you may want to look up CVT (continuously variable transmission) systems. There is one or two companies who have such gearboxes for bikes available (or they've showed such systems, not sure if they still sell them). It's probably what you have in mind. It's a mechanical system though, involving little to no electronics.
    – Slovakov
    Commented Apr 10, 2016 at 8:38

4 Answers 4


It could be done but would be expensive because you'd need torque sensing cranks. These exist for power monitoring but are priced as a tool for serious athletes. They would give you cadence as well. Then you'd need electronic shifting - also expensive.

The actual control could be implemented in any microcontroller of your choice, but you'd need to consider variability between pedal strokes (i.e. use a rolling average power which would make the system less responsive) .

In normal operation, you ease off a little for shifting (especially if you're working hard to start with). This would be impossible if the system shifted without you knowing. So gear changes would be harsh even if the system was well maintained (which you shouldn't assume).

If it gets a bit steeper and you're already in your bottom gear, you probably won't know, so while you're waiting for it to shift rather than standing on the pedals, you're losing headway.

Innovation in cycling tends to come from competitive disciplines, in which this idea would be a bad thing. Innovation in low-effort cycling does happen: e-bikes. But these are quite competitive on price, and there wouldn't be room in the market for an expensive add-on like this.

It could be interesting as an undergrad project in engineering, but I can't see it going further than that.

  • 1
    It would be simpler to have an electric-drive assisted bike, also called pedelec. The amount of power supplied by the electric drive is regulated by pedal power.
    – Carel
    Commented Apr 10, 2016 at 9:34
  • 1
    @Carel on any I've seen it's not regulated by power but by either (i) the action of pedalling with a switch attached to the cranks, or (ii) cadence/drive speed. Neither can tell how hard you're working. Also an electric assist motor takes a lot of battery compared to a sense/select system. Overall you're right though. Our first e-bike was bought from a lifelong cyclist who could no longer ride a normal bike.
    – Chris H
    Commented Apr 10, 2016 at 11:17
  • One thing I notice when I ride my 21 speed bike even here in very flat central Florida is even a slight change in load is noticeable and makes me want to shift and sometimes I bounce between gears multiple times. It would be nice if I could somehow program my bike to the load I prefer to pedal at and somehow it would not exceed that load (unless it runs out of gears). So then as I am pedaling along, it would adjust so that even if I go up hills, encounter headwinds..., the pedaling effort will never exceed the limit I set. Manual shifting does this by feel. Electronically would be nice.
    – David
    Commented Apr 10, 2016 at 11:20
  • The last thing I would want if the headwind gusted stronger would be for my gears to start messing around just as I have to put more effort in. Hills can be anticipated. You probably need to learn to be comfortable at a wide range of cadence (probably at the top end)
    – Chris H
    Commented Apr 10, 2016 at 11:25
  • People, they have torque wrenches that are not astronomically priced, so they should be able to have some torque sensing device on a bike which is "tied in" with a bike computer. It would be nice if the mechanical autoshifters also had an electronic mode so the rider could either select cadence based shifting or load based shifting or maybe a mode with both such that limits are set for both. Kinda like in a laboratory power supply where you have both current and voltage limiting, both of which can be set by the operator. Seems bike technology is lagging to me.
    – David
    Commented Apr 10, 2016 at 11:39

Shimano's Alfine Di2 electronic shifting system for hybrids has some ability to shift automatically, although I don't know if it is load adjustable.

  • 2
    I would assume that it senses cadence rather than effort (as cadence sensing is simpler).
    – Chris H
    Commented Apr 10, 2016 at 8:20

I don't believe such a device currently exists that you can mount on a bicycle, but there are devices that have characteristics similar to what you seek that you can mount a bicycle onto. They are specific types of bicycle trainers. The explanation is a bit long but if you understand how trainers work and how they compare to the work you do to move your bike outdoors on a real road you'll be better able to imagine what design challenges you'd face in creating the device you've imagined.

Outdoors on the road there are several sources of drag you must overcome. This bicycles.SE answer gives a fuller explanation but the short version is that there's a part of total drag that is constant with speed, a part that depends on your acceleration, a part that depends on climbing (or descending) a hill, and a part that depends on how aerodynamic you and your bike are. Importantly, the aerodynamic drag component doesn't scale linearly with your speed: it scales quadratically, so the the power needed to overcome aerodynamic drag varies with the cube of speed (assuming zero wind).

But the part that's related to your question is the "rolling resistance" component of drag. Rolling resistance is constant (that is, independent of speed) so the power needed to overcome rolling resistance is linear with speed. But the important part here is that the rolling resistance itself is constant. So if you are in a particular gear and you can ignore air resistance, your pedal load will be constant. If you speed up (and, since you're in a particular gear) your cadence will increase but the pedal load will be constant. This is the condition you were stating in your question.

Enter bicycle trainers. All bicycle trainers have a way to impose drag or to generate a load against which you pedal. You may be familiar with trainers that use turbines set into a fluid to create load, or wind trainers, or magnetic trainers. In general, the faster you pedal, the faster your rear wheel turns, which spins the trainer's load generator faster, which increases the load. Importantly, each trainer will have a different "load profile" which determines how its load increases with speed. Fluid and wind trainers (air can be considered a fluid) have the characteristic that their drag typically increases quadratically with wheel speed, like it does when you ride on a road outdoors. There are also bicycle trainers that act as ergometers; you set a particular target power and they adjust the load so that total power is maintained. If you pedal more slowly, the load increases to maintain targeted power load. If you pedal faster, the load decreases to maintain targeted power load. This is exactly opposite to what you want. However, there is a third type of training device: rollers. Typical rollers (that don't have any additional load generating device such as magnetic or fluid load units) have nearly constant drag -- which means that the power needed to overcome drag on rollers is nearly linear. So you're looking for a device that mimics rollers by adjusting the gear ratio to keep pedal loads constant.

What you seek is uncommon because most effort has been put into trying to make trainers less like rollers and more like riding on the road, where load increases nonlinearly. From an engineering point of view, it would be possible but from a marketing point of view, the demand works the other way.

  • Except rolling resistance does change even with constant speed. For example, I am riding on a sidewalk at 10 MPH then go onto the grass . If I maintain constant effort I will quickly slow down or even stop if I am in too high a gear.
    – David
    Commented Apr 10, 2016 at 22:37
  • Right, resistance also increases when you climb a hill at constant speed. This was a simple illustration of the design challenges you face. In fact, the bigger issue when you ride outdoors is acceleration (which you can see in that answer I linked to). In order to accelerate as you might do from a stop light, you have to increase load while still balancing on the bike. If you limit pedal load at a constant maximum your system will have to respond by gearing down very quickly. It's possible from an engineering standpoint but it will be hard to modulate it quickly enough.
    – R. Chung
    Commented Apr 11, 2016 at 1:03
  • An analogy I like to use is a constant current laboratory power supply unit (PSU) used to charge a battery. The battery will charge and want to resist the current flow so the lab PSU increases the voltage to keep things at a constant load (constant current). This is done so quickly that the ammeter never twitches from some set amperage (such as 1.500). Similarly, it would be nice if the bike was sophisticated enough to sense load changes and make the necessary adjustments in real time so from the driver's perspective, the load is unchanged (but the groundspeed will likely change).
    – David
    Commented Apr 11, 2016 at 1:15
  • From a full stop, the bike would simply shift to a low gear since it has to make some assumptions (unless it was really sophisticated and could sense when you were on a hill and how steep it was). It is somewhat amazing to me how simple it is for a bike rider to scan the scenery and feel the load and adjust manually and very quickly, yet when I ask a computer to do it all of a sudden it becomes a major project and expensive. The algorithm to handle a set maximum load (specified by the rider) would likely need refinements after riding and testing to handle all common circumstances well.
    – David
    Commented Apr 11, 2016 at 1:22
  • Right, in your design, from a full stop the bike would simply shift to a (very) low gear to limit pedal load. This is exactly the feeling when one first begins on rollers -- most people have far more difficulty learning to start up on rollers than to maintain speed on rollers because the resistance is so low when starting out. They fall over before they can balance. You've observed how resistance varies when you go from pavement onto the grass. In most cases, riders will "muscle up," not to go faster but to use that pedal resistance to attain balance. Your design removes that.
    – R. Chung
    Commented Apr 11, 2016 at 14:20

I think you could do this with a continuously variable transmission and some way to measure the load. You could DIY some pedals to measure with how much force the rider is pushing. The pedals would probably have to wirelessly transmit the data to a microcontroller so that you could adjust the gear ratio, since there isn't really a good way to run wires from spinning things to not spinning things. If you didn't need infinitely variable gear ratio, you could hook those pedal sensors up to a microcontroller and hack the system into some standard electronic shifting. It seems overcomplex.

Here is my actual recommendation. It may not give them the exact experience that they are looking for, but you're probably better off getting them an electric assist E-bike... or telling them to suck it up and get used to shifting.

  • The thing that sometimes annoys me about manual shifting is frequently having to "bounce" between 3 middle gears cuz of wind, slight incline... It would be nice if someone could just concentrate on the ride and have the electronics and mechanics wizardry worry about making minor adjustments so your maximum load setting is not exceeded.
    – David
    Commented Apr 11, 2016 at 15:34
  • I think something like an electric assist is exactly what you are looking for, rather than shifting gears and slowing down, the rider will get a bit of a boost to stick at the same speed. It'll probably be a more enjoyable experience all around.
    – Kevin
    Commented Apr 11, 2016 at 16:03
  • CVTs work in cars and scooters because they do much higher revs. A rider would do an utter maximum of 120-150 RPM before bouncing out of the saddle. By comparison, the input speed of a scooter is 2000-3000 RPM , and a car will do 6000-10,000 RPM. A CVT with an input speed of a hundred RPM isn't going to work the same.
    – Criggie
    Commented Apr 11, 2016 at 20:23
  • Sounds like fun. Maybe someday I will get a bike with a CVT such as the NuVinci 360 hub on a bike with those very wide tires so I can transition from sidewalk to grass and back to sidewalk without having the "clunkiness" of standard gears (chainrings and cogs). It will also give me a much lower underdrive that I desire and will eliminate ghost shifting which I dislike. I prefer a low gear bias since I rarely go more than about 15 MPH on my bike so I would like to gear it so it tops out at about 15 MPH at high cadence and then the 3.6:1 ratio will bottom out at about 4 MPH at high cadence.
    – David
    Commented Apr 12, 2016 at 2:13
  • Criggie. I don't think low RPMs of a bicycle is a problem for CVTs as it seems they can already handle that. I'd like to buy a used CVT someday to try it out. If anyone has an older model used CVT please let me know. I don't even mind if it is the heavy version like the older NuVincis. I just want to test it out, perhaps on a beach cruiser type bike frame which I already have in addition to my MTB which is a 3x7. I can just "shift" it manually at first then later I will consider that harmony electronic shifter. Thanks.
    – David
    Commented Apr 12, 2016 at 2:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.