Flame Seeking Boe-Bot Robot
![Andy Lindsay (Parallax)](https://forums.parallax.com/uploads/userpics/601/n66C2UR7NIHRG.jpg)
Introduction
Like the blue enhanced photodiode from Applied Sensors, light emitting diodes (LEDs) also conduct current at different rates with different levels of light exposure.· Since the currents are very small for ambient light levels, it turns out that the stray capacitance in the I/O pin, the metal clip under the breadboard row, and the LED's silicon junction store enough charge to make the RCTIME measurement work.· Try the attached circuit and code under controlled conditions.·
Circuit
·![attachment.php?attachmentid=51630](http://forums.parallax.com/attachment.php?attachmentid=51630)
Source Code
' TestIrLedFlameDetect.bs2
' See TestIrLedFlameDetect.PDF for Schematic.
·
' This is a novel application that uses the Boe-Bot
' robot's IR LEDs to detect flame in an area devoid
' of other sources of IR such as sunlight and halogen
' lamps.
' {$STAMP BS2}
' {$PBASIC 2.5}
time VAR Word
DO
· HIGH 8
· RCTIME 8, 1, time
· DEBUG HOME, CLREOL, ? time
· PAUSE 100
LOOP
Procedure
Make sure to keep the blinds drawn and eliminate other sources of IR such as incandescent lights.· Try the schematic and source code that begin with the word Test first, and make sure the Debug Terminal reports significantly lower measurements when a flame is shown to the dome on the LED.· Under these conditions, I've been able to discern an open (candle) flame up to a meter away.· Mind your cathodes and anodes too.· The IR LED is connected backwards from the way we usually connect it to emit IR for object detection.·
Your Turn
You can also try colored LEDs (red, green, yellow, blue) for detecting ambient light, or even use some colored LEDs·and IR LEDs to better discern flame from ambient light sources.
Post Edited By Moderator (Jessica Uelmen (Parallax)) : 8/25/2010 5:55:27 PM GMT
Like the blue enhanced photodiode from Applied Sensors, light emitting diodes (LEDs) also conduct current at different rates with different levels of light exposure.· Since the currents are very small for ambient light levels, it turns out that the stray capacitance in the I/O pin, the metal clip under the breadboard row, and the LED's silicon junction store enough charge to make the RCTIME measurement work.· Try the attached circuit and code under controlled conditions.·
Circuit
·
Source Code
' TestIrLedFlameDetect.bs2
' See TestIrLedFlameDetect.PDF for Schematic.
·
' This is a novel application that uses the Boe-Bot
' robot's IR LEDs to detect flame in an area devoid
' of other sources of IR such as sunlight and halogen
' lamps.
' {$STAMP BS2}
' {$PBASIC 2.5}
time VAR Word
DO
· HIGH 8
· RCTIME 8, 1, time
· DEBUG HOME, CLREOL, ? time
· PAUSE 100
LOOP
Procedure
Make sure to keep the blinds drawn and eliminate other sources of IR such as incandescent lights.· Try the schematic and source code that begin with the word Test first, and make sure the Debug Terminal reports significantly lower measurements when a flame is shown to the dome on the LED.· Under these conditions, I've been able to discern an open (candle) flame up to a meter away.· Mind your cathodes and anodes too.· The IR LED is connected backwards from the way we usually connect it to emit IR for object detection.·
Your Turn
You can also try colored LEDs (red, green, yellow, blue) for detecting ambient light, or even use some colored LEDs·and IR LEDs to better discern flame from ambient light sources.
Post Edited By Moderator (Jessica Uelmen (Parallax)) : 8/25/2010 5:55:27 PM GMT
Comments
Attached is a picture, zipped video clip, schematic, and source code of my flame following Boe-Bot robot. Although it requires controlled conditions and close range, it's still pretty cool that you can use stock Boe-Bot parts to detect and follow flame.
The measurements from the two sensors are pretty different, but they appear to be linearly related. At close range with the blinds drawn and only fluorescent lights in the room, it's doing well. A procedure for testing the relationship between a pair of given sensors and coding to compensate for the differences is on my to-do list. It appears to be linear over most ot the sensor range, which if true, will make it pretty easy to calibrate. That should also make it possible to direct an extinguishing device at, say, a candle.
Not sure if comparing the measurements to red LEDs and/or photoresistors can discern a flame under noisier conditions (with sun coming through the window and/or lights that generate a lot of heat in the room), but it seems worth a try. It would certainly be less expensive than other robotic flame detection devices folks are currently buying to participate in robotics firefighting contests.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Andy Lindsay
Education Department
Parallax, Inc.
Post Edited (Andy Lindsay (Parallax)) : 1/23/2008 5:47:31 AM GMT
I've seen this demo in Andy's office and it was impressive what he's demonstrated with a few dollars of parts. Another application for the Boe-Bot.
Ken Gracey
Parallax, Inc.
Cool project! Neat video too!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Whit+
"We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney
Post Edited (Whit) : 1/24/2008 2:26:56 AM GMT
Cool, uhhhhh, hot! Like a moth to a flame. Oops. One less Boe Bot.
LEDs make such good photodiodes, with low dark currrents due to the high bandgap voltage, I guess. In Applied Sensors, I'd consider using a plain yellow or green LED instead of the photodiode. (Partly because the price of that particular photodiode went sky high.)
A candle flame has a color temperature of around 1700 Kelvin, so there is not all that much energy at 940nm. But more than in the visible. The butane lighter would be a couple hundred degrees higher. Does it follow the light of a 940nm LED source? Maybe not. The emission spectrum might not match the response spectrum closely enough.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com
Can somebody post a sample for propeller using spin?
Thanks!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG
It worked pretty well, but I am sure it can be improved.
Thanks for the detection circuit and the other tip to get me started.
Wammo21