Un compañero y yo hicimos un proyecto similar con camara web puede pasar un pista similar a la tuya y ademas las decisiones no las toma con condiciones de colores sino que utiliza una red neuronal para el entrenamiento luego se implemento con xbee para no tener montada la lapto en el vehiculo jeje si quieres intercambiar conociento escribeme a mi cuenta :p
The programs on the right direction still, it dont need 50 reverse turns to actually get to the finish line. only if the laptop could turn its goddamn head to look around, it could easily overcome its fear of corners
@Souraneel Yeah but it "looked" before it turned where other robots just rammed the boundaries before readjusting their courses. Their robots are inefficient because there is the possibility of back tracking on a very tight "U" curve.
Awesome bot...is part of the problem in fluid movement caused by the video source being not absolutely real time? It seems that the source video for guidance lags a little and this may be an issue for fluid forward movement. Just wondering! Again AWESOME BOT! Love the fact you attached a notebook directly to this one! Far better for future programming upgrades or changes!
@DedicatedLearning As you said, the problem was the lag over the video detection (also, webcam image and framerate was very bad). The coding was done the last night in the hotel room, so we did not have the time to optimize it :(
The original championship track was very complex (with color signals and other stuff) but they finally decided to use this dumb track with just white borders. (A 'nuclear cannon' to kill flyes)
@soulreaverdoug hahaha es que pasamos de una velocidad tortuga llevando solo partes de cartón, a esa versión super veloz que llevaba bolsas de arroz sin inmutarse
Good implementation, hope you increase the speed next year.
I see that the robot moves with impulses. I think a PID can improve the speed. The error can be the difference between the actual angle of the robot and the angle of the curve of the road.
Well i would say you did not had a winning horse. but a nice try non the less.
I would say that the arduino in this basically was for servo control only. The main computations are made by the EEE which handles the video processing and then sends navigation commands through the usb to the arduino.
Anyway you had computational power to do way better. so you definitely need to improve on the video recognition side. Were you using OpenCV?
Which software do you use for camera snapshots? I had big latencies (>1sec) with every software I tried and it is unusable for realtime applications...
Un compañero y yo hicimos un proyecto similar con camara web puede pasar un pista similar a la tuya y ademas las decisiones no las toma con condiciones de colores sino que utiliza una red neuronal para el entrenamiento luego se implemento con xbee para no tener montada la lapto en el vehiculo jeje si quieres intercambiar conociento escribeme a mi cuenta :p
josar1787 6 months ago
nom nom nom
mrsp0re 9 months ago
The programs on the right direction still, it dont need 50 reverse turns to actually get to the finish line. only if the laptop could turn its goddamn head to look around, it could easily overcome its fear of corners
Souraneel 1 year ago
its really scared to take those crackin corners
Souraneel 1 year ago
@Souraneel hahah as we were!
eried 1 year ago
@Souraneel Yeah but it "looked" before it turned where other robots just rammed the boundaries before readjusting their courses. Their robots are inefficient because there is the possibility of back tracking on a very tight "U" curve.
MasterThief1324 1 year ago
Awesome bot...is part of the problem in fluid movement caused by the video source being not absolutely real time? It seems that the source video for guidance lags a little and this may be an issue for fluid forward movement. Just wondering! Again AWESOME BOT! Love the fact you attached a notebook directly to this one! Far better for future programming upgrades or changes!
DedicatedLearning 1 year ago
@DedicatedLearning As you said, the problem was the lag over the video detection (also, webcam image and framerate was very bad). The coding was done the last night in the hotel room, so we did not have the time to optimize it :(
The original championship track was very complex (with color signals and other stuff) but they finally decided to use this dumb track with just white borders. (A 'nuclear cannon' to kill flyes)
eried 1 year ago
va mas rapido q la mierda ............ lol
soulreaverdoug 1 year ago
@soulreaverdoug hahaha es que pasamos de una velocidad tortuga llevando solo partes de cartón, a esa versión super veloz que llevaba bolsas de arroz sin inmutarse
eried 1 year ago
MARIO KART SIM
alienjoe123 1 year ago
@alienjoe123 well, it can be remote controlled, so maybe it might be very funny to build a lot of these to compete with other people
eried 1 year ago
Nice last build!!!
Shame you had to tune it down so the cam could keep up :(
benatpc 1 year ago
@benatpc yes :( it was a last minute patch in the firmware controlling the motors (a delay to workaround the camera-road detection lag)
eried 1 year ago
constractulations!!!
nickmou 1 year ago
did you use the webcam on the netbook as a sensor?
ChipMunkMusicFan 1 year ago
@ChipMunkMusicFan yes
M2217 1 year ago
@M2217 Right!
eried 1 year ago
if there were a centerline award, you would've won it ;)
alexanderpas 1 year ago
@alexanderpas hahah :(
eried 1 year ago
Good implementation, hope you increase the speed next year.
I see that the robot moves with impulses. I think a PID can improve the speed. The error can be the difference between the actual angle of the robot and the angle of the curve of the road.
paucarre 1 year ago
Well i would say you did not had a winning horse. but a nice try non the less.
I would say that the arduino in this basically was for servo control only. The main computations are made by the EEE which handles the video processing and then sends navigation commands through the usb to the arduino.
Anyway you had computational power to do way better. so you definitely need to improve on the video recognition side. Were you using OpenCV?
assuncaP 2 years ago
It was opencv, but because the so many bugs of that library (sometimes just hangs) now another member of the team is exploring new solutions.
That time, we had electrical problems, our robot was a lot quicker than that :(
eried 2 years ago
Which software do you use for camera snapshots? I had big latencies (>1sec) with every software I tried and it is unusable for realtime applications...
slush0 2 years ago
A custom c# software for testing and the final version was in ansi c
eried 2 years ago
but which API? Some direct access or v4l?
slush0 2 years ago
@slush0 V4L2 and luvcview descompressing routines (it is UVC based)
M2217 1 year ago
hi very kool dude
did u use image processing for checking the lines or were using some line detection sensor.
If u were using image processing then how was arduino working arnd , were u doing d image processing in matlab or where xactly were u dng.
Can u drop u r email id
notgud20 2 years ago
Image processing was used for checking the lines. In this demo the processing is using C# (at first), but then was in C over Linux.
eried 2 years ago
Hi guys, the project is awesome!!!
I'm italian living in Taiwan for business in electronics fields, if you have any idea just let me know!!!
Anyway what song is the first one?
Is so cool!!!
rickyprose 3 years ago
Thanks for reply, the first song its "Ding Dingaling" from Vagabond.
eried 3 years ago
nice video, its awesome
congrats D:
paolonroll 3 years ago