Several builds implemented some version of this already, such as this
Sensor is a tiny stress sensor on the truck, its wired to the ESC, it measures deformation of the metal under load, no load=apply breaking, in case of failure ideally nothing happens, maybe an audio tone, your emergency brake just doesn’t work. Has nothing to do with pushing or going up hill, should only activate when there’s no rider.
I see you just want to do it without thinking about possibility of dying because of brakes failure. I’m out until you answer my questions and find a way to brake or understand that remoteless ride might kill you. You cannot jump off the board while going 50kmh downhill.
False. It just might hurt though. 
*safely 
I’m guessing that was implied 
@webst you wont die from a 15km/h crash normally… no one intends to build a 40km/h wireless board i think. I also could build a 80km/h hoverboard, and probably would end in death, just think before you do something
I missed your answer on how do you want the board to distinguish you pushing from self accelerating down the hill without remote? Do you guys want to have some kind of dance mat on your boards to take control of all possible situations? 
Normally that crash on tartan track should end pretty well: https://www.youtube.com/watch?v=bVZz8EN_NKw
Its been stated several times in the thread already.
“Keep in mind I’m not considering this for high speed boards, just low speed cruisers that go as fast as you can kick.”
Plenty of people go down hills on boards now with no breaks of any kind. But if that’s a concern, its a couple of lines of code to add a hard limit on max speed the board will let you reach, so that you won’t accelerate past your safe running speed.
I still don’t think that going limited 10 to 20kph with no emergency brakes is safe so I’m going to stay with the remote but you could just use available arduino remote code and tinker with it. It’s not that hard and communication with vesc is already done, add a button for starters, limit speed to 5kph and go from there.
Consider following scenarios:
Powerslide, beats every of your tree scenarios. There are people out there who have actually skatet before the invention of eskate, and we know more than one way to brake.
It definitely beats crowded pavement scenario aka human bowling 
If only you had some sort of friction pad say on the bottom of your foot that you could, I dunno. Drag? On the ground? Then no one would need brakes at all.
I guess if you put wireless temperature sensor in your friction pad it could actually inform your esc on your intention to brake else it would just ignore it. Not sure if this wouldn’t count as remote though.
I was under the impression you could run code directly on the vesc, why bother with arduino? What are the additional benefits?
Yes, you can run JavaScript on it. VESC has JavaScript Interpreter
/s
You can run code directly on the VESC as long as you want to compile firmware and open-source your changes if you sell it.
Doesn’t the vesc have access to rpm data from the motor (sensored motor)? If that’s that case the additional friction (from the foot breaking), under constant current, would cause a sudden drop in the rpm, which could trigger the vesc to reduce current to 0 and coast, allowing the rider to continue further slowing the board with his foot to a stop. Am I misunderstanding something?
Ommiting high powered, high torque setup that might not notice much, wouldn’t terrain elevation do the same? That is the problem I’m talking about from the beginning. The same is when you accelerate, every hill would accelerate you to vmax but I guess here we could connect it with the foot detection procedure(described further) and block acceleration. We could get sensor detecting that you moved your shoe on a side of the board connected with a mic data so the system would ignore most other objects passing by. That would engage brakes only in scenario when the foot is on the side of the board and noise levels (actual foot dragging) are above certain treshold. That should minimize false detections to certain degree and allow you to slow down while dowhill. Dragging your foot for 1s while downhill would stop acceleration, dragging longer would slow down deck until you disengage your feet or stop you completely when the foot is in braking mode. This unfortunately makes it pretty hard do turn and brake at the same time but then I’d also make rider detection that would stop the motor instantenoulsly when you’re out of the board and15kph vmax should be pretty safe to jump off. Accelerating would work in simillar manner only the foot movement would be detected by two sensors one after another that would cause rise in wheel rpm. This new higher rpm level would be your new high speed. There could be also an option to gradually slowing you down (more natural/Mellow style) or staying at the top speed until you start braking.
Sensors are to be selected, I’d go with rfid sensors with foot bracelet for added security but for testing any proximity sensor should do.
@webst I think key is to keep it simple. The last thing we want is add signal processing on an audio channel.
Originally, in order to differentiate uphill from foot drag I proposed tilt sensor, but you just gave me an even simpler idea.
Switch to coasting if BOTH conditions are met
This allows you to safely control speed via foot drag, shift your feet around while skating as much as you like and maintain power and still be able to cruise up a hill as long as you keep your back foot down on the stomp pad.
Additionally board comes to automatic stop if front AND back stomp pad have zero pressure for a full second.
Can anyone confirm that a VESC can in fact reliably detect additional load on the motors via drop in RPM??
I still think that you want to detect temperature difference in boiling kettle after putting your finger inside. It might work for certain temperature levels and some fairly underpowered kettle. Either way: not safe at all.