SVO: Fast Semi-Direct Monocular Visual Odometry (TRO'17, ICRA'14)

[@pedro_pinheiro]

Thank you for this amazing contribution to the community. Cheers from Brazil!

[@naikrovek]

Well this is the coolest thing I've seen all day.

[@Dalekmun2010Two]

I haven't been this impressed with tech in so long, this is utterly fascinating. All I can think of is imagine what it will do to VR, perhaps I'm just not creative enough, but still. Valve's lighthouse is wonderful as to how low demanding it is, but configuration in a diy setting's a bit iffy. If you can just stick a few high fps cameras on handheld controllers, that'd make DIY setups much easier.

[@offchan]

I know that it's been a long time since you commented this but I want to say that the problem is not about sticking a camera to the handheld controllers. The problem is about processing the algorithm inside the controller. You need chips for it which mean price and size.

[@pdvelez]

Great job! I have tried package using ROS Groovy and a PS3 eye camera (640x480@60fps). However, I haven't been able to get results similar to the first part of the video. The trajectory points are scattered and not continuous. Do you have any recommendations about the environment if it's indoors? Are there any other considerations for other cameras? Thank you

[@Ankit-ul1hc]

+Pedro Velez Hi Pedro Velez, I am also trying to run svo but it constantly keeps losing the features. I have posted the complete problem on github.com/uzh-rpg/rpg_svo/issues/134 It will be good if you could help me out. Thanks

[@h2o11h2o]

Hi, I have a question: Is this method sensitive of weather condition?

[@Arnoldo189]

When I trying to run the "live.launch" does not work to me. I have sent a e-mail to you, I hope you can help me.

[@DamianReloaded]

Impressive work! I can see this being used as a backup system for drones though I think it may have trouble with stormy/lighting weather conditions, even though it could be overcome by switching to other spectrums of light maybe..?

[@김수호-w2s]

admire you guys!

[@kaankaan4518]

fast make error on raspberrypi3 error: This file requires SSE2 support. Check your compiler flags. please help

[@ailabRPG]

Our latest work on fast semi-direct monocular visual odometry. It runs at 400 fps on a i7 laptop and 60 fps on an Odroid (e.g., Samsung phone). The video also shows its onboard application to autonomous flight of a quadrotor. kzbin.info/www/bejne/aIrReoCcrJuVgLs

[@xinguo2645]

ailabRPG Hi, I get really interested in visual odometry related work. This one really impressed me. However, I am nobody but a fresh man to this area. Can you please help me with the build steps. Specially, I would like to make this work on mobile device. Thanks

[@joemeno123]

may I know which camera are you using?

[@dtiydr]

+Juan Miguel Jimeno It would really matter just as long as it have quite good resolution to be able to detect and track details.

[@florentinalexandruiftimie8214]

Good guy ailabRPG, creates dope tech and open-source it.

[@nocnestudio]

Where to buy?

[@minds_eye123a]

wow this is amazing

[@PatrickTresset]

Very impressive.

[@bbogart99]

I suppose similar approach as those generating position from accelerometer data, using something like optical flow?

[@chrislzy4959]

nice project

[@jacksonkr_]

I know they are monochromatic but could this work with a night vision camera? This giving "wings" to night time flying.

[@karanchawla8458]

Yeah you can use an IR camera and use this for VO

[@enriquejimeneztrejo590]

I would like to learn more

[@HassanOmariprofile]

amazing