Very clever! However, I have a question: if there are multiple routs to the centre and it learns both, how does it judge which to go down? If they both have the same amount of turns, how does it work out which is best?
Hrm. Well done! I think I'll give this a shot. I have done line followers (AVR-based) and maze solvers (a pure PC software in Java), it would be fun to merge them:P
hei urs so simple and very efficient. i also did 1 follower its not much efficient as yours and not able to get your speed at al.and how much big you codded for this
What does the robot do if it can't complete the first pass in the time alloted? The original rules for line maze (using 1/4" tape instead of 3/4") made it very difficult to complete a first pass in only one run.
@tzkelley The rules for the competition this robot was designed for didn't have time restrictions that made it difficult to learn the maze in a single run, so it wasn't designed to perform a multi-stage search. I have since programmed a 3pi maze-solver that uses a flood-fill algorithm to solve looped mazes. With this version of the code, the robot can store the entire explored state of the maze in EEPROM as it goes, so it could explore over multiple runs if programmed to do so.
We do sell most of the 3pi parts by themselves if you want to build something similar from scratch, and you can add custom electronics to the 3pi.
You can make a line-follower with one sensor if you really want to; adding more sensors gives more resolution for PID and allows for better performance. The two outside sensors (1 and 5) are good for detecting intersections in mazes or seeing things off to the side of the line course. Line-following can be done well with just the 3 inner sensors.
Youtube comments are not the best medium for discussing software improvements. Please take this discussion to the 3pi robot section of the Pololu forum.
is 3pi the fastest maze solver in the world. is there any other you have seen? Can u refer me to. i am getting 3pi in india for compition , i just want to be sure. Can u send this program , i will clean it.
The most I can say is that the 3pi is the fastest commercially available maze-solver that I'm aware of. The code from this video is not publicly available and would not help you much given its complexity and how finely tuned it is to this maze's grid spacing and surface. You should use the sample maze-solving code we provide in the Pololu AVR library and work on modifying that to be faster. Please use the Pololu forum if you need suggestions on how to do this.
Most of the components are listed as related products on the 3pi product page. You can also look at the page for the Pololu 5" round robot chassis for parts ideas (product #s 1500 - 1507).
The program is a custom maze-solving program that isn't in the Pololu AVR library examples. It's finely tuned for the particular course material and grid-spacing and the code is fairly complicated and messy (it was written in one night from scratch), so it wouldn't be all that useful as a general example. I recommend using the 3pi-mazesolver example as a starting point and then incrementally modify it to improve speed and performance. Use the Pololu forum if you need help.
Hi! Nice robot! I am interested in purchasing one of these, but don't know how complicated the programming will be... Could you please post a sample of the program you used for this video? Thank you!
Thanks for your interest. Please take a look at section 8 (Example #2: Maze Solving) of the 3pi's user's guide, which you can find under the resources tab of the 3pi product page linked in the video's description. Hopefully this will give you an idea of what is involved in programming a 3pi maze-solver.
You can buy the 3pi robot from the web page in this video's description. It is a programmable robot, so it does what you program it to do. We offer libraries that make it easy to use the on-board hardware, and you can download sample C programs from the product web page for line following and maze solving. These sample programs intentionally leave plenty of room for improvement so you can have the fun of figuring out how to enhance the robot's performance.
The sensors are phototransistor/emitter pairs that work by measuring the amount of light that is reflected back to the detectors from the IR LED emitters, so they are only able to distinguish surfaces based on differences in their IR reflectances.
No, this particular algorithm is designed for non-looped mazes. I plan to modify it so that the next version can handle mazes with loops (i.e. multiple paths).
good day.
i'm really amazed with your maze robot project.
if you don't mind i would like also to make one.
may i ask for the circuit diagram of your project?
please send it to angelies1291@yahoo.com
thank you.
geliegelie 2 weeks ago
what will happen if you let it remeber this course and put it on an other course?
jipje44 2 weeks ago
This video went viral on Chisinau
nickstout513 1 month ago
hello my friend, what motor you used in this project?
engalcm 1 month ago
THAT WAS AMAZINGGGGGGGGGGG!!!!!!!
thesuperjake2 3 months ago
one question can you some how a make a case for the robot so his body parts dont show up. other than that... GENIUS
gamefreak615 3 months ago
i'm aMAZEd!
Qwedvit 4 months ago
Very clever! However, I have a question: if there are multiple routs to the centre and it learns both, how does it judge which to go down? If they both have the same amount of turns, how does it work out which is best?
Scoonezy132 5 months ago
if you notice, robot just using the left wall rule.
He just turn left, when it possible. That's all
sorry 4 my english, i'm just russian nerd :D
MeikaKosake 5 months ago
Hrm. Well done! I think I'll give this a shot. I have done line followers (AVR-based) and maze solvers (a pure PC software in Java), it would be fun to merge them:P
ps: Great victory song! :P
hhrm 5 months ago
the victory song ^^,
raymondgreat 6 months ago
Hello, please tell me what's the width of the black line. Thanks!
yo9hnhradio 8 months ago
I see your algorithm... thanks a lot dude
anigamusuk02 8 months ago
cute little bugger!
tyxas 9 months ago
how it record the the route ?
please tell me
rendygaul 10 months ago
hei urs so simple and very efficient. i also did 1 follower its not much efficient as yours and not able to get your speed at al.and how much big you codded for this
dinezeazy 11 months ago
wow!
Guiga280498 11 months ago
nice map
003exia 11 months ago
amaziiiiiiiiiiiiiing :D
monarosebarrientos 1 year ago
it's too bad it's not compatible with ti-basic.
ZILOGz80VIDEOS 1 year ago
AMAZING !!!
airsp 1 year ago
What does the robot do if it can't complete the first pass in the time alloted? The original rules for line maze (using 1/4" tape instead of 3/4") made it very difficult to complete a first pass in only one run.
tzkelley 1 year ago
@tzkelley The rules for the competition this robot was designed for didn't have time restrictions that made it difficult to learn the maze in a single run, so it wasn't designed to perform a multi-stage search. I have since programmed a 3pi maze-solver that uses a flood-fill algorithm to solve looped mazes. With this version of the code, the robot can store the entire explored state of the maze in EEPROM as it goes, so it could explore over multiple runs if programmed to do so.
BenPololu 1 year ago
awesome !!!!!!
mragw 1 year ago
I feel tempted to buy one, but i prefer making my own robots from scratch...
why are 5 QTR-RC reflectance sensors used?
i made a line follower using only 3, and i've seen them with only 2 working fine
96bikerider96 1 year ago
We do sell most of the 3pi parts by themselves if you want to build something similar from scratch, and you can add custom electronics to the 3pi.
You can make a line-follower with one sensor if you really want to; adding more sensors gives more resolution for PID and allows for better performance. The two outside sensors (1 and 5) are good for detecting intersections in mazes or seeing things off to the side of the line course. Line-following can be done well with just the 3 inner sensors.
BenPololu 1 year ago
Thats cool! how did you make that?
Dmob911 1 year ago
i am trying code 3pi for a line maze.there is 6 inch difference and no loops. what modifications to sample code you suggest or it is the best ?
karandex 1 year ago
Youtube comments are not the best medium for discussing software improvements. Please take this discussion to the 3pi robot section of the Pololu forum.
BenPololu 1 year ago
what % of speed increase from stock program in both run.
karandex 1 year ago
is 3pi the fastest maze solver in the world. is there any other you have seen? Can u refer me to. i am getting 3pi in india for compition , i just want to be sure. Can u send this program , i will clean it.
karandex 1 year ago
The most I can say is that the 3pi is the fastest commercially available maze-solver that I'm aware of. The code from this video is not publicly available and would not help you much given its complexity and how finely tuned it is to this maze's grid spacing and surface. You should use the sample maze-solving code we provide in the Pololu AVR library and work on modifying that to be faster. Please use the Pololu forum if you need suggestions on how to do this.
BenPololu 1 year ago
if i want to replicate 3pi in some way what parts from pololu i need ?
karandex 1 year ago
Most of the components are listed as related products on the 3pi product page. You can also look at the page for the Pololu 5" round robot chassis for parts ideas (product #s 1500 - 1507).
BenPololu 1 year ago
i saw that your program is not stock one. Please send me this program .or just give hint.
karandex 1 year ago
what program did you use from the pololu-avr/examples files
flipit4199 2 years ago
The program is a custom maze-solving program that isn't in the Pololu AVR library examples. It's finely tuned for the particular course material and grid-spacing and the code is fairly complicated and messy (it was written in one night from scratch), so it wouldn't be all that useful as a general example. I recommend using the 3pi-mazesolver example as a starting point and then incrementally modify it to improve speed and performance. Use the Pololu forum if you need help.
BenPololu 2 years ago
that robot doesnt look ahead when solving mazes
cappstv 2 years ago
nice project!
BulKathos84 2 years ago
that so nice......
mpus06 2 years ago
That is a very nice robot. Nice design!
adnanfofo 3 years ago
Hi! Nice robot! I am interested in purchasing one of these, but don't know how complicated the programming will be... Could you please post a sample of the program you used for this video? Thank you!
annieconde11 3 years ago
Thanks for your interest. Please take a look at section 8 (Example #2: Maze Solving) of the 3pi's user's guide, which you can find under the resources tab of the 3pi product page linked in the video's description. Hopefully this will give you an idea of what is involved in programming a 3pi maze-solver.
BenPololu 3 years ago
How would I get one of those. And do they all do that or do you have to program it.
Abrakadeadguy 3 years ago
Also it would be great for searching lost people in a Haunted maze....(sigh it happend to me)
Abrakadeadguy 3 years ago
You can buy the 3pi robot from the web page in this video's description. It is a programmable robot, so it does what you program it to do. We offer libraries that make it easy to use the on-board hardware, and you can download sample C programs from the product web page for line following and maze solving. These sample programs intentionally leave plenty of room for improvement so you can have the fun of figuring out how to enhance the robot's performance.
BenPololu 3 years ago
Nice robot, it can only detect light luminosity or detects well other colors?
one2one121s 3 years ago
The sensors are phototransistor/emitter pairs that work by measuring the amount of light that is reflected back to the detectors from the IR LED emitters, so they are only able to distinguish surfaces based on differences in their IR reflectances.
BenPololu 3 years ago
impressive speed, but does your algorythm if there is more than one possible path?
sammyF70 3 years ago
No, this particular algorithm is designed for non-looped mazes. I plan to modify it so that the next version can handle mazes with loops (i.e. multiple paths).
BenPololu 3 years ago
That is a very nice robot. Nice design!
JeromeDemers 3 years ago
wow it's down right fast
blacklocist 3 years ago
very impressed. Love it!!!
WizmoMagic 3 years ago
impressive :)
creave 3 years ago
that's pretty amazing. i'm impressed.
stryker213 3 years ago