Drewby: Robot

Microsoft Robotics Studio

Tue, 20 Jun 2006 13:54:43 GMT

Ok, this is cool. I heard some whispering about this, but had no idea what it was. Now you can check it out! It looks like Microsoft Robotics Studio lets you program your robot in C#, VB.NET or even Iron Python and the test it in a simulation environment.

Microsoft Robotics Studio

Microsoft announces the Microsoft Robotics Studio, which offers end-to-end robotics development environment customer technical preview for hobbyist, academic, and commercial developers.

Robot Update

Thu, 18 May 2006 04:18:19 GMT

I’m long overdue to write an update on my Windows CE-based robot, powered by .NET. I’ve been heads down in getting the robot to work before the Day of .NET event in Ann Arbor, MI last week. I’m happy to report that I did “complete” the robot although it did not work during the presentation at the event due to a Platform Builder issue.

Here’s a rundown of the major events leading up to the completion:

Shortly after presenting a Platform Builder session to developers in Detroit a few months ago, I started having an issue with my Gumstix. Basically, I would upload my Windows CE image once, everything would work, but after that session, the Gumstix would no longer boot. I sent it back once for reflash ($10 + shipping) thinking it was inadvertent, since I had loaded images several times already. However, when I got the Gumstix back, I loaded an image and the same thing occurred (yes, $10 + shipping again). Since then, I’ve rebuilt and sysgened my OS and have not seen the issue again.
Meanwhile, I started assembling hardware. I had my chassis, gearbox, motors and motor controller hooked up to the serial port of my laptop. I wrote a quick .NET application to control the motors with sliders. That was a cool milestone for myself, albeit no big deal for hardware experts.
I ordered an ADC chip (TI TLV2543) and got that communicating between my sensors and the PXA255 (processor on the Gumstix). I figured out later that I could just connect my sensors (which are all off/on at the moment) to GPIO pins and receive the ADC for true analog sensors (such as temperature, pressure, etc).
I'm using two sensors. One is a line tracking sensor, which is actually three seperate IR sensors facing downward. The other is a forward looking IR sensor used to avoid objects.
I decided to add some push buttons and LEDs to the robot so I could select different programs to run and get some feedback from my code.
I connected everything up and disconnected the umbilical cord (the network cable between my computer and the Gumstix) from the robot and watched it work autonomously for the first time, less than 24 hours before the Ann Arbor Day of .NET. Talk about just in time.

As far as programming the robot, the WE-DIG guys have done an excellent job with the Drumstix project by adding some managed libraries for directly manipulated registers. This means I can access functionality of the PXA255, including accessing GPIO pins. Writing to a register is a simple single call to the Drumstix managed IP:

IO.Register.Write(address, value)

I created some libraries for the robot as well. For example, I created a class to represent a hardware button. This allows me to instantiate a button object, passing in the GPIO pin number that the button is connected to, and register an event handler for when the button is pressed. For example:

_button1 = new Button(GPIO_BUTTON1);
_button1.Pressed += new EventHandler(button_Pressed);

I implemented a similar class for LEDs connected to a GPIO pin:

_statusLed = new Led(GPIO_LED);
_statusLed.TurnOn();
_statusLed.TurnOff();
_statusLed.Flash(1000, 5, true);

The class I use for accessing the ADC exposes each analog input on the ADC via an indexer. So that code is fairly simple:

uint left = _adc[ADC_AIN_LEFT];
uint center = _adc[ADC_AIN_CENTER];
uint right = _adc[ADC_AIN_RIGHT];

Taking input from the ADC, I can then make decisions about which motors to turn on and off. The motor controller is accessed via TTL line, connected to COM1 on my device. I wrapped the serial communication in a class to make it easier to control the motors. So for a left turn, I stop the left motor and start the right motor:

_motors.SetMotorSpeed(MOTOR_LEFT, MotorDirection.Forward, 0);
_motors.SetMotorSpeed(MOTOR_RIGHT, MotorDirection.Forward, 30);

Currently, the robot still has lots of issues. Two months ago, I was sitting at my desk with a breadboard and an electronics 101 book trying to get an LED to turn on and off (“oh, the long lead is positive!”). So most of my issues right now are in hardware. For example, I don’t have the right batteries (9 volt drives the motors for about 10-15 minutes). There is also some weird behaviours when not connected to the computer (which makes it tough to debug).

Lots of people ask me about project costs. I believe I’m at about $250 for the robot. Most of that is sunk in the Gumstix device ($100) and Expansion boards ($150). The rest is in the robot platform, motor controller, sensors and other electronic components. It doesn’t really matter. I’m having SO much fun with this thing and it rocks to write C# code to drive the robot!

More to come...

Windows CE Based Robot

Thu, 30 Mar 2006 15:14:16 GMT

I've been tinkering with embedding Windows CE on a GumStix platform, which I'm detailing in another series of posts on Platform Builder. But I needed something to do with this device. After considering different projects, I decided a robot would be fun and there are plenty of resources on the web to help me along.

I don't have any experience in electronics, so it will be an interesting challenge. The main components of building the robot (the chassis, motors, gears and motor controller) seem to be straight forward, but I still haven't figured out exactly what I'll do for analog sensors, an Analog/Digital convertor, and logic to read the ADC. The audiostix I purchased with my GumStix appears to have some of this functionality in the same chip (UCB1400), but I'm not sure if I can use it yet.

I chose a robot platform and components that I think I can get working, given the lack of experience I have. At the same time, I wanted to make sure they left enough of a challenge to make the project interesting. I didn't want it to be a complete kit. I also wanted to cost to be low enough that if I fail, it wouldn't hurt too bad. I found the parts at http://www.pololu.com, but you'll probably see similar kits all over the web. Here's what I ordered:

Round Robot Chassis - A cut piece of plastic on which everything gets mounted.
Chassis Completion Kit - Twin-motor gearbox, ball caster, truck tires
Dual Serial Motor Controller - Also available in a pre-assembled micro-sized version, but I wanted to solder something.
5" Round Prototyping Board - To create circuits for my sensors and other components.
1.25 inch Male/Female StandOff Set - To mount the protyping board to the robot.

I also recommend you get at Breadboard and Jumper Wire Kit if you want to experiment and learn how to build different circuits before you solder them into place. The breadboards at Pololu are a LOT cheaper than at Radio Shack.

Yesterday, all the parts arrived and I started assembling the robot. I'll detail that in another post.