I would like to know if the propeller is the right tool for my project. Many t

jmbertoncelli

Hello,
I am new here and I do not have a lot of embedded programming and electronic skills but I think that I am a good programmer with more than 25 years of expertise with platforms like IBM mainframes, Windows, Linux,...
My wife has MS and she cannot walk anymore and her arms are weak, she has·an electric wheelchair but our house was not designed for a person with disability.· In fact she can navigate from one room to another but she has a lot of trouble when the space to navigate is not large enough... In fact there·is enough space to make the move but the motion control must be very precise and that is why she cannot make it.· I decided to help her as much as possible and I got an idea.· My idea is to add an embedded motion controller that could help her to drive the wheelchair when the available space around the chair required a more precise motion control.· I have already made a small prototype using the LEGO Mindstorm NXT but I have now to decide what could be the best target embedded microcontroller platform for this project.· I shall be able to control motor(s) speed and position, IR and US sensors for objects detection (virtual bumpers), Light sensor(s) "yes, I want to follow the 'line' to make the room to room navigation less complex", RFID reader to read tag on the floor "line(s) junction information to make the·room to room navigation less complex", a Compass sensor and a Tilt sensor.· I do not know yet if I can just build on top of her wheelchair or if I have to build a new one from scratch...
·
I would like to know if you think that my project is feasible based on the Propeller(s) or·if there is something better that I haven't yet found or think about?
·
Many thanks for your support and help.
·
jm.

PS: I am French and I apologize for my poor English.

rjo_

jm

I am an end user... with no special relationship to Parallax. I am not an engineer.

Everything you listed is pretty much available for the Propeller... There is no close second to the Propeller. I am completely biased, but I started with an open mind about a year ago.

The Spin language is entirely intuitive... very easy to pick up. The compiler is not completely described, so you sometimes have to ask yourself... "what?"

The only problem I see is directly controlling the main drive motor... the HB-25 controller is fabulous but peaks out at 14 volts.

It sounds like your wife is having fine motor control problems... you might consider trying to solve the problem by reworking the hand controls. She probably has problems remembering what is behind her(don't we all?)... so a couple of cheap cameras strategically placed might also help.

To get started I would suggest ordering both the education kit and the demo board. The demo board is a "hoot."

As you look through the object exchange and also the forum, you will find that most of the technical issues that you mentioned are fully discussed and that working code is available in several different forms.

It sounds like you will need a good accelerometer... that's also available!

I don't think you will find anything better than the Propeller... and unlike working in the mainframe environment, the half-life for what you learn about the Propeller is very, very long. There is no risk of learning something today... and then having to learn something else tomorrow.

A Propeller or a series of Propellers can do just about anything... and there are plenty of folks ready to help you here.

Rich

mcstar

Wow, this is a very interesting project you have come up with. I was playing with this not too long ago, but for robot navigation. My solution took me deep into study about dead reckoning, path navigation, motor control, etc. Many years ago I worked in a Motorolla manuf. plant that had a couple robots that could run around the plant delivering parts to the assembly lines. They didn't use anything other than line following, IR and bumper sensors. Their memories contained a map of the entire facility ad the allowable paths they could use to got from place to place. This is similar to the type of setup you'll need.

You will need to create maps of the entire home. Think of it as a 2D grid of the entire location. These maps are just 2D definitions of the walls in the home. Something like Wall - {x0,y0} - {x1,y1} where x0 and y0 define the start of a line defining the wall and x1,y1 and the other end of the line defining an edge of the wall. Then you could have locations defined like this {xk,yk} where xk,yk are the position of the center of the kitchen for instance and {xl, yl} are the location of the living room. Finally you'd created paths, just lines that define the allowable or safe paths between locations. Using this information you can create algorithms that can navigate a device between any two locations in your home. Search for information on Dead Reckoning and path finding for more information, or post questions here as you get into it. You can use IR reflectors in known locations to calibrate your positions to compensate for inevitable drift that ocurrs when travelling over carpet and other surfaces . You'll need poisiton sensors on the two main drive wheels so you can caclulate the chair's position as it is moved from place to place. Once you have a setup that can accurately locate itself inside your home, you can start looking into path finding algorithms. You can have "optimum paths" setup in the map of the home. A rather simple algorithm can determine how much to move each wheel to close the gap between the current location and the desired path once a desired location to move to is selected (from list of known locations for example).

Electronically, you'll need to design a way to interface your prop with the input control system of the chair. I think they just use variable resistors on the joystick. I work for a company that sells scooters and powerchairs, so if you tell me the model/manufacturer of your wife's chair I may be able to get schematics for you.

Skogsgurra

An alternative to "follow the line" is to have reflectors (with a crude "personality") and a rotating scanning device that understands the difference between the reflectors. Rest is as per previous post.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔

jmbertoncelli

Hello, I would like to thank you all for your quick answers. I shall keep this post up to date and let you know about my progress I know know that if I have questions I can get your support!
Again thank you all!
jm.

Fred Hawkins

Suggestion: use proximity sensors (and a propeller) to manage the speed of the motors -- close means slow. That way your wife retains the initiative, and the wheelchair keeps moving until it actually hits something at a hopefully very slow speed. Eventually you and she ought to find the optimal slowdown that lets her retain control.

The obvious advantage is that you don't need to map anything. And the wheelchair can work anywhere.

Lawson

you might also try to give the wheel chair a springy repulsion from any nearby walls. something like (virtual force away from the wall) = K/(distance to wall). With walls to the left and right this should cause the wheel chair to track down the center of a hall. Hm, doors would need a special case. Both so the chair didn't go shooting into an open door along a tight hallway, and to look ahead to center the chair with any oncoming doorways. Also, narrow legged chairs and tables along a hallway would really screw this up. I think MIT or CMU build a wheel chair like this in the 90s.

My idea would also work well with Fred's idea. The bonus is that once debugged this combination should work anywhere. On the flip side, setting up "train tracks" with black lines on the floor in the tight areas of the house should be super simple to get working. (with the right line sensor the black lines for the "train tracks" could also be on the ceiling)

Since a person is controlling this thing at all times, navigation and path finding really aren't needed. Just leverage the smarts of the operator for these functions. Also, if the operator has an override switch, they can easily save a dumb but useful system from getting stuck. (for instance, if a line follower crosses from carpet to hard floor, it might see the edge as a line to follow! with an override switch the operator can get it back on track quickly)

Looking forward to seeing the final product!
Marty

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Lunch cures all problems! have you had lunch?

W9GFO

I think you'll have to build a new chair, or at least get a new chair and rip out the motor controller and replace with your own. I picked up a chair recently to use for a tracked robot project. Originally I thought that I could just intercept the motor control signals from the joystick and recreate them using a Basic Stamp or propeller. On my chair the joystick and controller are one unit. Originally I thought that the joystick unit would send control signals down to a motor controller in the base but that's not the way they work. The joystick, programming and motor control are all in that one unit.