I love this project! It would be interesting to see the processing requirements for the use of the model (not training). Putting the inference on a dedicated microcontroller would be pretty cool.
This is an extreme project you are working on! I really enjoy watching these types of autonomous vehicles. They motivate me a lot. Keep up the good work!
I am working on doing the same thing. Its very cool that today we have so much technology that can drive based on cameras instead of bumper switches now. Your project turned out great and gives me inspiration to keep working on mine. Good job
Well done, great build!! I'm impressed by your passion for mechanics and fabrication (I see some pretty nice machines in your workshop) AND autonomous driving technologies. Thanks for posting!
Not so long ago? Man, not to be a hater but when that video posted I was going through a motorcycle training course. Since then I've bought a motorcycle and put 600 miles on it over weekends. I was excited to see what you've gotten up to in that time. Thanks for the update.
Of course, it’s completely your decision, and I respect whatever you decide. Either way, your work is inspiring, and I’m grateful for what you’ve shared so far!
It might be cool if throttle input was one of the training data variables. Even if the self driving model doesn’t have direct control over the throttle, it might help it to alter it driving style at higher speeds if it knew how fast it was going.
If possible, get the vehicle to follow a line of tape in the middle of track. That would minimise learning confusion as the tape would always be on camera.
What did you use to put arbitrary images and animations on an oscilloscope? I have a few programs that put their own stuff up there (like a clock or a demo song), but I don't know how to tell it "show this image".
@@austiwawa Always one more thing to do. I hesitated about that though because the training might end up teaching it to go off course on purpose. Not sure how you could teach an AI that something is wrong and how to handle it but avoid the scenario in the first place.
remember as a kid when you wanted a go-kart? now remember how you fought with your siblings to have a go when you got one? to the point your parents took it away and locked it up until "you learnt to share"? did you ever stop and think "gee, wouldnt it be awesome if it drove itself so i could sit here and just watch?" yeah... cant say i ever did that either.
You reproduced the DonkeyCar project. Training with a constant speed helps a lot. The reaction of the steering angle will become consistent when the speed is constant.
It should be illegal for someone to be this smart, cool, talented, dedicated, hardworking AND handsome 😁😍👏🏼 this is the coolest project yet!!!
Riz
Can't wait for the autonomous drifting video
Dude, incredible. Bring this to open sauce and let people ride it in the power wheels track! lol
Thanks, Sean! I would actually love to LOL
I've always wondered about doing this at RC scale. Very cool to see the small sized rideable experiments!
I love this project! It would be interesting to see the processing requirements for the use of the model (not training). Putting the inference on a dedicated microcontroller would be pretty cool.
This is an extreme project you are working on! I really enjoy watching these types of autonomous vehicles. They motivate me a lot. Keep up the good work!
Thanks a lot!
What a great project and thanks for the insides
Thanks!
I am working on doing the same thing. Its very cool that today we have so much technology that can drive based on cameras instead of bumper switches now. Your project turned out great and gives me inspiration to keep working on mine. Good job
Thanks and good luck with your project!
Amazing! Keep up the great work Austin!
Thanks Jay!
Awesome work man! Love that you’re always troubleshooting and never give up!
Well done, great build!! I'm impressed by your passion for mechanics and fabrication (I see some pretty nice machines in your workshop) AND autonomous driving technologies. Thanks for posting!
Thanks for the awesome comment!
Not so long ago? Man, not to be a hater but when that video posted I was going through a motorcycle training course. Since then I've bought a motorcycle and put 600 miles on it over weekends.
I was excited to see what you've gotten up to in that time. Thanks for the update.
I guess it has been over a year now xD
Of course, it’s completely your decision, and I respect whatever you decide. Either way, your work is inspiring, and I’m grateful for what you’ve shared so far!
Thanks! I appreciate the support!
Brilliant, Austin! Really fantastic work!!! 😃
Stay safe there with your family! 🖖😊
Thanks MC!
12:20 Not a huge fan of raspberry pi's but I feel like this would be a good place for one.
It might be cool if throttle input was one of the training data variables. Even if the self driving model doesn’t have direct control over the throttle, it might help it to alter it driving style at higher speeds if it knew how fast it was going.
I should definitely include throttle data in training
12:54 - You didn't fool me. 😂
If possible, get the vehicle to follow a line of tape in the middle of track. That would minimise learning confusion as the tape would always be on camera.
This was going to be my backup plan haha
This is awesome!
Thanks!
What did you use to put arbitrary images and animations on an oscilloscope? I have a few programs that put their own stuff up there (like a clock or a demo song), but I don't know how to tell it "show this image".
osci-render :)
You should add throttle to the mix and train it on leaving the track by backing up and slowing down on the sharp corners.
Yes, I should definitely include throttle in the training!
@@austiwawa Always one more thing to do. I hesitated about that though because the training might end up teaching it to go off course on purpose. Not sure how you could teach an AI that something is wrong and how to handle it but avoid the scenario in the first place.
12:10 that is why you whould never ever connect cirectly to the speed input without anything inbetween.
Yes, I was risking it without an e-stop lol.
@@austiwawa Keep the videos coming, they are super entertaining and inspiring
@@nrdesign1991 Thanks. That means a lot!
The future is here. Excellent work!
Thank you Hyuk!
MY GUY!!🔥
Hey Ben!!
@austiwawa proud of you Austin! Keep doing your thing man!
remember as a kid when you wanted a go-kart?
now remember how you fought with your siblings to have a go when you got one? to the point your parents took it away and locked it up until "you learnt to share"?
did you ever stop and think "gee, wouldnt it be awesome if it drove itself so i could sit here and just watch?"
yeah... cant say i ever did that either.
Another great video!!
Thanks Isaac!
So cool work, thank you
Thanks for watching!
I wonder if a ring light behind the camera would have helped or made it worse with the detection of yellow and black tape.
Interesting idea!
I am trying same course but still unable to attech steering with potentiometer. Can you please describe it briefly
You reproduced the DonkeyCar project. Training with a constant speed helps a lot. The reaction of the steering angle will become consistent when the speed is constant.
Thanks for the tip. I should have made the throttle circuit sooner and used it for training!
My cat chewy would love a car like this for our apartment, cool video 10/10 !
Thank you!
where are the views!? This is Sherlock Holmes level detective work, Great engineering.
Hopefully more will come! Thanks a lot!
This is incredible. How far or much more difficult would be to make the machine stop when it identifies people or other cars?
Thanks! Object detection is a feature that I would like to add but I have not experimented with it yet.
Thanks Austin
Thanks for watching!
Love you bro from India 🇮🇳❤
Jay shree Ram well done nice project 🙏
Thank you!
Nice
any files for the project available ?
Not at the moment. But I am thinking about sharing them.
You should have a mechanical bumper switch that cuts power......
🥳🥳
Did you ask Tesla if they need a partner?
singsong narration noooooooo! 😓