What You Can Do With NVIDIA Jetson Nano?

anyway we proved we've got this uh elephant thing what else can we try to identify what we've got here this is a uh it's an orange that's an orangey elephant still in the shot go away elephant it's always wants to get in on the action that elephant let's try um switching fruit let's have a that's a granny smith very higher confidence on granny smith looking at 99 points something i think it was 100 then for a second 100 percent now if it says the granny smith has become close to the camera so it's it's good at recognizing not just apples particular types of apples and presumably is it good at say yes it's very good at the banana pretty certain that's a banana it's amazing ready it's picking this thing out amid the background of my hand and everything else isn't it [Music] so i've written some really basic code that just brings those gpio pins high and low depending on which image it sees so basically which category is so a square makes it stop so we'll leave that there and the triangle makes it go and that's really just to test out my motors and check i've got everything wired correctly and basically the algorithm that recognizes the images is giving me data my if statements work and then i can actually control the wheels but we want to make something a bit more complicated than that what i love about it is that it is kids it is silicon valley engineers it is retirees people who aren't engineers at all this event is uh an opportunity for our community to show their stuff and these are small size and autonomous cars using the various different techniques and the history of car innovation we've always been through racing the only exception is we multiply the speed by 1.2 just because that sounds cool and at the beginning we hit the gas really hard for these courses where you're trying to collect data typically what we do with these guys pretty much driving around the course collecting images and throttle and steering information it's doing that 20 times per second as the car is traveling down the track so with the extra power that the jetson provides you can actually bump that up up to about 60 frames per second capturing that data and that information is very crucial when you're making these tight turns on this track there's what it's seeing you can see whether it's a blocked or unblocked let's put that on the screen so you can see when it thinks it's blocked it will turn when it isn't blocked like now it won't turn put my foot in or it'll turn there we are and i'm impressed with this now clearly this could be trained to actually be more accurate in this particular area but it does actually work it had to see is moving around detecting what it can see here i am back again and as you can see it's working the robot has learned how to move around across the whole surface of this table without falling off [Music] hey [Music] you

Channel: Cytron Technologies

Views: 61,360

Keywords: Jetson Nano, Artificial Intelligent, Neural Network

Id: NYM0l1JZIVA

Length: 4min 17sec (257 seconds)

Published: Tue Apr 06 2021