Compound Eye

@karpathy @StewartalsopIII 2006: "method hallucinates a large occupied area in the center of the road ... [our] approach overcomes this error and generates a map that is good enough for driving." robots.stanford.edu/papers/thrun.s…

@StewartalsopIII I think it might have been me in my 2015 RNN blog post 🤦‍♂️😅. It’s the earliest mention I could find

Hey @karpathy do you know who came up with the term hallucinations?

Great article in @Automotive_News about our new VIDAS Development Kit: autonews.com/mobility-repor… Find out more on our website and contact us for a live demo. We'll mount our system on vehicle of your choice and drive any route, on-road or off! compoundeye.com/vidas-devkit

Saw this today. Amazing computer vision. ⁦@compoundeyeinc⁩ Another stealth mode company that does amazing things with just two cameras. Will talk about this tonight on ⁦@ThisIsMeIn360VR⁩ Twitter Space that goes from about 6 p.m. to midnight. x.com/compoundeyeinc…

Watching the recording of the @TEDchris interview with Elon Musk at #TED2022 on a Saturday afternoon and l can’t help but wonder if @elonmusk has seen a demo of @compoundeyeinc !? compoundeye.com

There are two myths fueling the lidar vs camera debate. Neither are true. Here is a quick overview of our myth busting episode with @TheAutonocast: bit.ly/3KxFEZE

Help us build the next generation of autonomous vehicles. Join a team of experts from Apple, Google, Argo, SRI, and Mapbox like @TaraniDuncan

Join our team of researchers and scientists from companies like SRI, Argo AI, Apple, and Google. See open roles here. compoundeye.com/careers

This IIHS survey rewards close reading. Two camera-only systems beat three (but not all) camera-plus-radar systems. The software can be more important than the sensors. iihs.org/news/detail/pe…

Using 6 stereo pairs of cameras for 360D obstacle detection, @JohnDeere generates raw image, depth map, per pixel classification, identifying ground, sky, crop, and foreign objects that may fall into a field. bit.ly/3h4oNks

Great to see @ConsumerReports' Kelly Funkhouser on @CBSMornings giving an accurate overview of the current state of driver assistance technology. youtube.com/watch?v=W3n6bh…

An amazing presentation by camera lead at @PonyAI_tech on imager selection. bit.ly/3sZGrva

Join our team of researchers and scientists from companies like SRI, Argo AI, Apple, and Google. See open roles here. compoundeye.com/careers

We had a blast on @TheAutonocast! Find out why there’s more to perception than “one camera or lidar”

@JasonTorchinsky This summary of the SAE levels by @PhilKoopman shows there are many things people get wrong. A safety system that makes a car uncrashable could still be L0! Driver supervision is optional for L2! L3 systems do not have to warn the driver to take over! users.ece.cmu.edu/~koopman/j3016/

It's amazing how many people who criticize any take on Tesla Autopilot or other L2 systems don't really seem to understand the SAE level system at all. jalopnik.com/lots-of-people…

Our first blog post gives an overview of how our camera-only 3D perception system sees the world compoundeye.com/post/introduci…

@greentheonly This is a complete system - automotive cameras of our own design plus software - it doesn't work with arbitrary video. But we are streaming live demos from our test vehicles every day, so you can check those out.

@compoundeyeinc Where do I upload my own sample for a test?

.@greentheonly You might be interested in our depth estimation from video: stream.compoundeye.com

What we’ve been working on: real time, long range, high resolution, 3D perception using only cameras. stream.compoundeye.com

#3D mapping. #ISR. #PNT. We need #innovation and these #SmallBusinesses are bringing it to xTechSearch 5 #technology pitches: @ColdQuanta, @compoundeyeinc, @ElphelCamera, #Enview, @Hivemapper, #HLAcquisition, @IDenTVSF, , @polarissensor, and #UniversityTechnicalServices.

Proud to be a part of the IJRR 2018 Robotics Research paper of the year! With Mustafa Mukadam, Jing Dong, Xinyan Yan, and Byron Boots.