Self-driving cars are something we are continually told will be the Next Big Thing. It’s nothing new, we’ve seen several decades of periodic demonstrations of the technology as it has evolved. Now we have real prototype cars on real roads rather than test tracks, and though they are billion-dollar research vehicles from organisations with deep pockets and a long view it is starting to seem that this is a technology we have a real chance of seeing at a consumer level.
A self-driving car may seem as though it is beyond the abilities of a Hackaday reader, but while it might be difficult to produce safe collision avoidance of a full-sized car on public roads it’s certainly not impossible to produce something with a little more modest capabilities. [Jaimyn Mayer] and [Kendrick Tan] have done just that, creating a self-driving R/C car that can follow a complex road pattern without human intervention.
Unexpectedly they have eschewed the many ARM-based boards as the brains of the unit, instead going for an Intel NUC mini-PC powered by a Core i5 as the brains of the unit. It’s powered by a laptop battery bank, and takes input from a webcam. Direction and throttle can be computed by the NUC and sent to an Arduino which handles the car control. There is also a radio control channel allowing the car to be switched from autonomous to human controlled to emergency stop modes.
They go into detail on the polarizing and neutral density filters they used with their webcam, something that may make interesting reading for anyone interested in machine vision. All their code is open source, and can be found linked from their write-up. Meanwhile the video below the break shows their machine on their test circuit, completing it with varying levels of success.
We’ve featured quite a few self-driving cars over the years, and this isn’t the first model-sized one. Among others we’ve seen a self-driving Acura built in a garage, and an autonomous vehicle course from a Stanford professor.
21 thoughts on “Self-Driving R/C Car Uses An Intel NUC”
Please someone educate me as to why this isn’t a glorified line follower, because it looks like a glorified line follower.
Agreed. Let’s see it drive the same path at night. In the rain. With oncoming traffic.
With my mother-in-law complaining in the backseat.
That’s what all self-driving cars actually are – glorified line followers.
The google cars get their line from the initial scanning run by a human driver, which is subsequently cleaned up and annotated by people and then uploaded back to the car to repeat the journey. The Tesla cars are just following lane markers and road edges – it’s a glorified cruise control / lane watch that Tesla falsely markets as an autopilot.
That’s all human drivers are, too.
HUMANS FALSELY MARKETED AS DRIVERS
Well, you could be ignoring the lanes or driving in the wrong way if you’re suicidal.
For the “self-driving” car to do that takes a wet leaf in the camera and a flaky GPS connection.
Well, it doesn’t *need* a line, it looks like it should work with any kind of landmarks.
Hokay then, as soon as we get the road crews out to paint big fat black stripes down each side of the lane we’re all good then.
That would be easy to do here in Montreal, Quebec, each and every highway have traffic cones permanently installed
Hm, self driving cars. Had a thought about that a short while ago, while I was leaving a busy Walmart. Want to start a little simpler? Well, come up with a car that can park itself in a large busy parking lot. So you drive right up to the front door, get out, and press a button on your remote. You go in and shop while the car parks itself. When you’re about ready to leave you click it again, and the car comes find you at the front door. No more searching for a empty spot within walking range, no more dashing through the rain. And no more jerks parking at an angle and taking up two or three spots.
Should be a piece of cake compared to a busy freeway and fast moving idiots, er, distracted drivers (yeah, that’s what I meant to say, fast moving distracted drivers. That’s my story!)
I’m sure people are about to tell me why this is a bad idea, but that’s okay. Some people just be like that…. Lol.
Unfortunately you are more likely to hit a walking person or an open door that appears usually without warning. Hit a truck at 5 MPH usually everyone survives, hit a person at 5 MPH… :(
For now it will be parking-assist only where you still can slam on the brakes or turn the wheel. Government rules say someone needs to be in the drivers seat.
But I’m looking forward to fully automatic parking as you described in the future for those who would truly benifit.
P.S.: “…no more jerks parking at an angle and taking up two or three spots.”
That’s not going away anytime soon! :P
There will be an “asshole parking” button on the remote.
Hahaha, I love you! :P
I think the goal is great, and possibly more useful to a majority of drivers than fancy cruise control on freeways. Possible complications; there are a lot more people walking around inside a parking lot than in a street or freeway, and signage is not standardized.
I was going to say, ahh but there won’t be any people in the parking lot, we get out at the entrance and our cars park themselves…… and other positive sounding stuff…….
but then I saw the video. So now I’m stuck with arrrrg we’re all gonna die!
Learning to drive is just like coloring in. You just need to stay between the lines
Machines don’t have advanced pattern recognition and decades of cognitive training. You could run genetic algorithms in a test lab using a video game rendering to a video frame and letting the car learn to avoid collisions for 10 years. Eventually it will encounter every possible scenario and have a safe reaction outcome to every collision event.
What if terrists let it play carmageddon?
Yes I agree with you self driving car uses an intel nuc
Self-driving succeed during the passage of time and you have explained through videos it perfect scenario.
Hey, one of the developers here. For some context we were both first year university students who built this in our free time as a fun side project. All up we probably only spent a couple of weeks so I think we made good progress. ;)
@RW I guess you could say it’s a glorified line follower but it still used a neural network that took some weeks to build, train and test. It would also work on any other landmarks like cones, tape, etc. Roads happen to use lines to mark lanes so we opted to use them.
@mjrippe if we had enough data it would definitely drive during rain and at night. It gets very repetitive while training it so we figured we’d just collect daylight data. You get quite dizzy after walking in circles for 20-30 minutes :)
Please be kind and respectful to help make the comments section excellent. (Comment Policy)