There's not a whole lot of energy to recover -- most of the eBike's energy is spent fighting wind resistance, so there's not much left to recover when you're slowing to a stop at a stop sign.

Long descents where there is significant energy to recover are not that common for most riders.

A stop should provide almost as much energy as starting back up again. How about an e-bike with a fixed gear? You get assist when starting, but pedal yourself at speed, regen to stop, and perhaps some option to charge the battery via the pedals that's up to the rider.

The most unusual thing I ever saw in my EV days... We had a full Ford Focus Electric on a chassis dyno to debug some problems. After several hours of tuning and data collection we realized the car needed to be charged to get it back to our own facility. The dyno place had no high power chargers. So we set the dyno somewhere around 40mph, put the car in drive and left it. With no foot on the gas it would do a light regeneration at that speed. The strange part is that we sat around a table to analyze the days data about 5 feet from the car. We're at a table next to a car tied down and going 40mph and its almost dead silent charging via regen. Analysis done, car charged, untie it and off we went ;-)

Define "almost as much energy". Bike motors aren't super efficient, especially at low RPM's (like when you're coasting to a stop).

A bike has very little rolling resistance, nearly all of the energy used to propel the bike on level ground is going toward fighting air resistance. Even on rolling hills, I rarely use my brakes on the downhill, any regen braking used on the downhill is just going to increase the energy needed to go up the next hill. (and due to less than perfect efficiency, you lose energy on both the regen capture and powering the motor again, so it's a big loss).

So yes, you might recover some small amount of energy used to decelerate the bike at a stop sign, it's not going to be significant percentage of the total power requirements, and you're going to be paying a price in dollars, complexity and weight for this feature.

Well you could just use the drive as an electromagnetic brake w/o necessarily recovering all power by diverting all or just a part of it to a load. Train engines do this.

Well you could, but it comes at a cost (in dollars, complexity and weight), and is likely not worth it since the bike still needs friction brakes.

It's a different story with a 50 ton train where a 20 KW resistor bank is not a significant amount of weight, but adding a 500W resistor bank to a bike (since you need somewhere for energy to go when the battery is fully charged) is probably not going to be worth the weight and expense.

Plus you need some sort of blending between regenerative and friction braking, so it's yet another thing that needs adjusted and maintained.

Even a DC motor driven with an H-bridge can dump mechanical energy back into the battery. It's trivial.

My Bionx Xtracycle has regenerative breaking and it is as implied super-useful as a speed moderator here in SF when going downhill with a load (my two kids) would build up speed.

It's firmware limited to 20 mph which _feels_ slow but is probably a sane compromise. We only hit that going down hill, and are far from the fastest cycle; but we average 12-15 mph on mild uphills and flats when conventional cycles usually average a few mph slower.

I love my ebike, it makes a daily commute taking two kids to school then getting on to work feasible, regardless of weather or relative personal energy level.

Towards the bicycle end of the spectrum you probably want it to be able to freewheel.

I expect it is a cost/benefit thing, where an ebike is already quite pricey, and equipped with a very reliable fallback.