Robot project - useing stock automation items

[maroe]

Hi. I'm a service engineer trainee (in automation) from Norway.
A friend om mine, and I are going to make an balance robot on two wheels. I have seen many robots on the internett like this, but all of them are made of electronic parts. We are going to make one with stock automation items.

THE BODY
The body of the robot is made up of 5 floors.
1. floor: Tilt sensor - FAS-G (gyroscope and accelerometer)
2. floor: PLC (programmable logic controller) - CPU, encoder innput, pulse output, input, output, analog innput, analog output.
3. floor: Solid state relays - For pulse-width modulation, and motor fase switch.
4. floor: Power pack - A 24VDC battery
5. floor: Compact metal for extra weight. (and connection space for switches)
top: Control panel - Switches for ON/OFF, PROGRAM MODE, WIRELESS CONTROL


This is how far we are today. The PLC mounted is not the one we'll be using, just mounted it for show.


The 1. floor is made of thin steel, and the 2.-5. (and the top) is made of plexiglass (a special type plexiglass that is stronger, and do not brake that easily)
Under the 1. floor, there are 2 DC motors (with gearbox), and 2 motor encoders. The DC motors are attached to an 14mm shaft that goes trough a bearing.
The encoders is connected the the shaft with a strap or a cog wheel (not sure which we'll go for).


THE LOGIC
We will use the PLC (OMRON CJ1M) for regulating the balance, driveing-command (first by pre-made programs in the PLC, and later by either wireless control, line-following or sound following), and all other stuff like buzzers, lights, and so on.
The only exception is pulse-width modulation, and motor fase switch, which will be done by solid-state relays. (5pic x 2)



I have no real point of posting this, but I tough you might like to read it, and I welcome comments.

[fr4ncium]

Some students at my university made a Segway a few years ago.  Some other students and I were going to make an updated model and were testing out the old one - someone hit it into a wall hard last year and some of the stuff became uncalibrated.  Just a warning - you are not going to have a fun time making the algorithm for calculating when the robot is upright - the step size inaccuracies in the gyro and accelerometer are going to build up over time and eventually, what it thinks is upright might not be and vice versa.  Just a heads up.

And, as you have already done in the current version, keep the center of mass as close to the wheels as possible. 

Here is a website you might find useful, although building a Segway-like machine and a self-balancing robot are two very different things.
http://www.tlb.org/

[Kohanbash]

What is the total mass?
Have you decided yet how you are going to filter the output of the microstrain?

[maroe]

What is the total mass?

I don't know that yet, because we have not built the thing yet, only the framework.
I can let you know when we have all the parts 
(and we aare building the framework in out workshop at work, and we have nothing to mesure weight with  )

Have you decided yet how you are going to filter the output of the microstrain?

As I understand, the FAS-G has a built-in filter?
Please tell me more about this....

Ps. this is infact our first robot, and I guess this is not the BEST type of robot to start with. I would not have started with something like this at home on my own, but this is a work project, so lets hope we aren't that over our heads 

[maroe]

At the moment I got some time on my hands (my boss has to approve the costs of all the components, and then I can order parts ), so I was thinking about robot control (or AI if you will).

This is what type of control I have been thinking of:

- Sound control --> Use 4 microphone that gives analog signals to the PLC, the PLC is programed to only read the two microphone-inputs with the highest value (the 4 microphones is placed at each side of the robot (N, W, S, E) and by only reading the two with the highest value, we know the 45 degree angle.). Then we measure the total value of the two, to find the distance (This is of course depend on that the sound source gives the same amount of sound every time). Then we calculate the value percentage of the two, to get the exact angle.

- Line following --> That I think is so common around here that I don't bother explaining

- Visual control --> Using machine vision. Run a vision program on a computer, have a wireless camera and transmitter on the robot, and let the program on the computer tell the PLC how to move. Like a Vanishing Point program using Roborelm or a similar vision program.
And if we'll use visual control, we will most likely also make a direct control over the robot through a drive interface on the computer (maby through the internett?)

-Sonar beacons (mapping) --> If I make the robot capable of reading sonar signals, I can set up 3 sonar beacons around it's work area. Then I can find out the accurate position of the robot. And when he knows his position, and if he knows the positions of things he are going to interact with, he can operate independently 


any thought? 

[Kohanbash]

Since the FAS-G is designed for dynamic applications (and will be used in one)The FAS-G's filter is pretty good if you are using the RS-232. In analog mode you will probably get noisy data back that you might have to average a couple of values.

[maroe]

Since the FAS-G is designed for dynamic applications (and will be used in one)The FAS-G's filter is pretty good if you are using the RS-232. In analog mode you will probably get noisy data back that you might have to average a couple of values.

Ok, I didn't realize that, thanks! 
Yes, then I'll be useing the RS-232, with the PLC card CJ1W-SCU21-V1. That sould work perfectly 

The PLC I will be useing:

OMRON

CJ1W-PD022    -    Power Supply
CJ1M-CPU22    -    CPU, 2 x encoder input, 2 x pulse output (the 2 pulse output will control the motors trough solid state relays)
CJ1W-SCU21-V1    -    2 x RS-232 port, Protokoll Makro
CJ1W-ID211 (SL)    -    16 x digital input
CJ1W-OD212 (SL)    -    16 x digital output

[maroe]

First parts has arrived 
The two motors and the solid state relays.
I made a motor control with another PLC I found on work, with the SSR's. Worked perfectly 
The rest for the parts will likely arrive next week, except the FAS-G, it has a delivery time of a couple of weeks 

Now that I know the motor control works, it's time to start mounting the motors, wheels, encoders, and stuff like that.
I'll come back with updates (and questions if needed )