PDA

View Full Version : EduRobot autonomy


dlavery
12-01-2002, 02:18 PM
Has anyone started to really exercize the autonomy aspects of the EduRobotics kits yet?

There are a bunch of exercises designed for the EduRobotics kit on the NASA Robotics Education Project Robot Autonomy Activities (http://robots.nasa.gov/edu/lessons/autonomy.htm) page. The programming subteam on Team 116 has started to work through several of them, to get some experience with the autonomy capabilities of the kits.

One of the first things we tried was using simple photoresistors as cheap optical sensors, and developing a line-following capability. You can see some of the results here (http://www.invisiblerobot.com/robotics/line-follow/). By adding in two cheap Sharp proximity sensors (do you notice a recurring theme of "cheap" running through many of our posts? :) ), we also built a version of the mini-robot that maintains a 4-inch standoff distance from any approaching object. You can see a Quicktime movie of it here (http://www.team116.org/2003/images/RepelRobot.mov). Each of these examples took less than one afternoon to program and implement.

What are other teams doing with the kits? Let's see if we can get a good discussion on EduRobot autonomy going! (and maybe share some code?)

-dave

Ken Loyd
12-01-2002, 10:07 PM
Dave,

I have not yet had the opportunity to jump into the Edurobot (our sister school has it for now). The reason for this reply is to ask you if your are aware that there is a RoboEducators Conference in Las Vegas on December 14th at the MGM. I realize this is short notice, but if you can make it, "Come on Down!". I will be attending and can send you the highlights if you would like.

Ken Loyd
Team 64

dlavery
12-01-2002, 11:42 PM
ken -

I've had the Dec 14 meeting on my calendar for about two months. I am going to try to make it, but the trip may get bumped by other travel conflicts. Probability is about 50% right now.

-dave

seanwitte
12-02-2002, 09:08 AM
Dave mentioned the sensors we used, here is some more information. I'm not an expert, but these seem to work pretty well. Please let me know if there are any errors.

IR Rangefinder - Sharp GP2D120, about $13.00
I bought them at Digi-Key, but realized that they don't come with the JST connectors. If you get them from Acroname.com they include the connectors, which will save you some time. You can put a regular Futaba J-Type 3 pin connector on the other end and plug it directly into an analog input on the RC. Make sure that the output pin is connected to the center pin on the connector. It outputs a voltage between 0 and 3V that correlates to the distance detected, up to about 30 cm. Check the data sheets.

Visible Light Phototransistors - Jameco part #120221, product number BPW77. Available at jameco.com, about $0.99.
These are nice because they have a built-in lens that limits the focus to about 10 degrees in front of the transistor. You pretty much have to point it straight at a light to get a reading. I'll try to get Kyle and Byonge to put together a white paper on it. I heard about them from a web site called robotroom.com, which has lots of good ideas. To use it, cut off the base pin and connect +5V to the collector, a 100K potentiometer to the emitter, and the pot to ground. Take the output voltage from the emitter/pot connection. The brighter the light, the higher the output voltage. For the line sensors, we added a bright yellow LED to light up the area below the transistor. Tweak the potentiometer value to see how you can adjust the sensitivity. Higher values detect very small changes, lower values large changes.

IR Emitter/Detector pairs - Radio Shack, about $2.50
We used these as touch sensors by measuring the amount of infrared light reflected back at the detector. These work exactly like the visible light phototransistors, but without the nifty lens. You'll need to put some electrical tape or heat shrink around the detector to focus it forward. We used a comparator chip to set a reference voltage, which was compared to the outputs from the IR detectors. The comparator output is either 1 (the detector voltage is above the reference) or 0 (the detector voltage is below the reference). By tweaking the reference you can make them more/less sensitive. We had them "detect" a white box from about 4" away reliably. The outputs were passed into the RC as binary inputs.

The circuit below allows you to connect external logic signals (0 - 5V) to a binary input on the RC. When the input is logic high (5V) the transistor connects the signal pin from the RC to the ground pin on the RC. When the input voltage is 0, the transistor disconnects them. The resistor is there to limit the current drain. This is called an open-collector switch.

---------------- Signal Pin from RC
| (left-hand pin, I think)
| Collector
10K Resistor /
Input------|====|--------------| NPN Switching Transistor
(0-5V) \
v Emitter
|
GRD ----------------------------- Ground Pin from RC
(from external logic circuit) (right-hand pin, I think)


The comparator is a bit tricky because the outputs are either disconnected or grounded. To make it logic 1 and 0, add a 10K pull-up resistor between the output pin and +5V. That will make it logic high and logic 0.

We have an external solderless breadboard with its own power supply (regulator IC from radio shack that converts +9V battery down to +5V).