I am trying to build an intelligent mountain board. The aim is to use computer vision to make intelligent navigation, meaning to assist the driver such as automatic positioning on the road.
It should provide with voice recognition for tweaking the drive parameters onboard. Plus, the ability to control the turn with a game-pad joystick.
So I started by just making it move. The current version is the bare minimum for commuting.
Currently, I am using quite weak electronics. A Chinese mepo like ESC with some old recycled 18650 cells in a 10s4p battery. In a few days, I upgrade to 18650 cells 11s12p pack since my area is mostly uphills and downhills(Denmark).
I am waiting for my odrive to arrive so I can upgrade to 13s12p and have quite good torque/speed. Once I get the odrive I will implement my python script with the odrive to be able to control turns with a joystick.
So far I have been playing with google API for voice recognition and seems like a good solution for voice control, but not sure how it will handle the outside noise.
As by me, I finished my degree in robotics and doing master in computer vision, so I hope to get a lot of computer vision work with the board since it is a vehicle. I plan on adding an IMU sensor and raspberry pi with some cameras and a GPS. These should allow building some intelligence.
P.s: Hopefully, I can manage to get it self balancing on the back wheels and follow function since it’s quite heavy and I would rather have it follow me by itself.
They’re actually not bad considering what they cost and bluetooth 5.0 audio quality isn’t the greatest but they’re loud and you can easily still hear your surroundings at reasonable levels
That would be mechanically difficult, but possible. Mosty the joystick control is by changing the speed at which each motor is rotating. I would love to have omnidirectional wheels and drive on 4 hub motors, but can’t see the omnidirectional wheels handling such a speed.
I have a wireless headphones with an inbuilt microphone, but i also need a helmet, so they don’t fit. I think at the begging it will a earphone with an separate microphone.
Put a platform on a linear slide on the deck with a stepper drive, to turn it moves the ‘deck’ to the side to lean the board without an active rider. Probably easy enough to toss a 9 axis and PID loop if you are running a central processor for navigation
It could be that the front truck is customized so it moves freely on the Z-axis and a stepper motor is used for rotation, so the user still can tilt the deck. That should be for sure something to try, but for that, I would most likely need a CNC component since the 3D printed part is not strong enough. My university doesn’t allow us to CNC components that are not required directly for a university project, so getting one might be tought
But to tilt the deck directly, might be a solution, I have a 3d printer and a few stepper motors with the leadscrew. So maybe 3D printing a component and controlling the tilt could be done in a few months.
I think the first is to get o-drive implemented has it has a higher priority. I am just not quite sure how responsive will be the change of trajectory of the board by regulating each motor speed.
So I will look into it based on how well the board will turn with o-drive.
@mmaner, can we bump @Iuliu to member? I invited him from the other side.
Welcome @luliu I have thought about making a skateboard that will just follow me around. As for steering, the first thing that came to my mind is a differential drive too, but wheel wear would be too much in a 4 wheel setup. Ideal would be what @ZachTetra said to shift weight around, right near the front wheels.
Got an update. I deiced to go with an 11s10p for now, since my meepo ESC works up to 45V.
My battery pack is made of the best 18650 recycled cells i could find. It’s a game changer then going uphill. In 4km, got from 45V to 44.5V. The speed is also quite good, i guess it’s between 25-30km/h, but the breaking got worse.
And for the end, i plan to add my original wheels on the sides and remove the rubber wheel. The wheels should be attached to the motor directly, or i should find bearings that fit both the wheel and the original bar.
since it’s quite heavy and I would rather have it follow me by itself.
