Equip Matterport Camera with LIDAR to give it more capability to work outside

I posted this idea just before the upgrade to the community forum occurred and it seems to have disappeared from the forum so here it is again...

The ability to scan outside with these cameras cannot be overstated.

Trying to scan in the early morning or after sundown "may" work but it's not always possible or convenient to do this and usually requires multiple visits to a site.

Furthermore, if you need to scan across a large open area that does not have any vertical objects along the path the system will generally not work. It simply does not have enough 3D objects to use to work out its new position.

Fortunately there's been a lot of development in the LIDAR world and the price of the technology has dropped significantly:

https://techcrunch.com/2017/04/13/luminar-debuts-a-sensor-to-make-self-driving-cars-safer-than-human/

https://techcrunch.com/2016/12/13/velodynes-latest-solid-state-lidar-design-keeps-costs-low-for-production-at-scale/

Combining this technology with the existing camera would be a huge game changer.

I'm quite certain that it would attract a lot of interest from clients in different market segments.

Please like this post if you agree! 

:-)

All the best,

Stuart

 

 

 

 

 

5

Comments

12 comments
  • Hi David,

    The article about Velodyne's latest solid state sensor should make this a "no brainer":

    "... Velodyne has a new LiDAR sensor design that could help bring an autonomous-driving future closer than ever; it’s a small, solid-sate design that has a cost of less than $50 per system (when purchased at the kind of high-volume rates needed for mass manufacturing). LiDAR’s cost has been a limiting factor in seeing it applied more broadly as part of semi-autonomous driver assistance systems and fully autonomous cars targeting future production."

    Even if Matterport couldn't manufacture in the volumes that Velodyne are looking for and the sensors cost $1000 each it would still be worth doing in my opinion. :-)

  • @David Hothersall thank you for providing the link.

  • To be clear - I just want a more reliable way of scanning outdoors than the"structured light" system that the Matterport camera currently uses.

    Buying another system and then trying to combine the different data sets together is just asking for trouble and outside the average users comfort zone I'm guessing.

    I would simply prefer that Matterport replace the existing structured light sensors with Lidar (or something else?).

    The Scanse system looks interesting and seems to compute the distances to targets using time of flight of its laser. I'm guessing that it has a high accuracy clock similar to this one:

    https://www.microsemi.com/products/timing-synchronization-systems/embedded-timing-solutions/components/sa-45s-chip-scale-atomic-clock

    • ±5.0E-11 accuracy at shipment  (1 to 2 cm accuracy)

    The good news is that this technology is rapidly dropping in price too.

    As I'm sure you'll agree - a lot of clients want the models to be able to go "outside" in order to see the exteriors of their buildings.

    The current system just can't be counted upon to do this unfortunately and I find that we often waste a lot of time trying to get it to work.

    The 360 degree panoramic views can help the situation but are often not enough to keep the clients happy.

  • Easy scanning outdoors would be amazing. We are always looking into new technologies like the ones you mention above.

    -Maddy

  • I agree this capability would open up the market for Matterport to an unprecedented level. This would move the device from interesting to essential in nearly every business that can make use of high quality scanning and imagery.

    However the temptation to move the pricing too far upmarket needs to be resisted. Most LIDAR devices are very expensive and even the most recent 'cheap' device like the LASiris from NCTech is around $12k but still doesn't produce 3D meshed models. So Matterport still have the lead there.

    Time taken to produce scans as well seems to be higher on LIDAR devices and the 30 second capture of the Matterport camera is also a great advantage.

    The range of LIDAR devices also would be a positive as it potentially reduces the number of scans needed, certainly in large buildings making the capture process even quicker.

    So combine the technology with a 'sensible' >$7k-$8k price and maintain the speed of scanning and that's the lower half of the market nailed in my opinion.

    I'm sure this wouldn't be for everyone but it's certainly a great opportunity if it can be pulled off.

  • @SSquires are you aware of the Scanse project? http://scanse.io/  Could be used with a ground robot or drone.  What's your take?

  • Thanks for sharing that info about the Scanse project Lisa!

    The price looks pretty reasonable - I'll definitely have a look at it.

    Hopefully MP can integrate this or some other technology into the cameras - it would definitely make the MSPs lives easier! :-)

  • @SSquires ... I have worked with Volopervidi as they wanted to potentially integrate Matterport Scans with their Lidar work; accuracy of the Matterport was not what they needed on their projects. Personally I was only viewing from a great way to integrate the outside information with the Matterport which does not require accuracy per se. 

    Derek Bohannan with Volopervidi sent me the following - his personal thoughts (below) and I've reached out to Scanse to see about integration with MP scans.

    From this at 1000cm or roughly 32.8 feet you can have measurement variability of up to 1.64 feet. Scanners we are using scan 100+ feet and we are getting errors in the hundredths of an inch. I would have to look at what applications are asking for LiDAR data that isn’t required to be extremely accurate. You can get much more accurate by overlapping the scans if you have good target matching in a target rich environment but then what does that do to your efficiency?

     

  • I've looked at the scanse project quite closely and while its  nice and is a pretty attractive price I don't think it produces cloudpoints any more accurate than the Matterport. 

    Have a look on their forums for real world results as a few guys have posted their scans on there. There are a few errors in some of the scans due to the assembly of the device causing a 'corkscrew' effect as it rotates but I guess it does win with regards to working in all lighting conditions and over longer distances.

    http://community.scanse.io/t/3d-scan-examples/331

    There are of course no image files produced so the point count isn't colourised.

    Worth keeping an eye on st the price point though.

     

     

  • I too would prefer a Matteport solution.  *crossed fingers*

  • I've mentioned this to Matterport on several occasions in emails and responses to surveys, so I suggest others take the opportunity when they occur. I've also joined the AEC group and the guys tasked with driving the tech into that space are aware of my interest so I'd say again its worth making them aware of your interest.

    Sensibly though, I think we'd probably be looking at a Pro + camera. The existing structured light scanning is fine for the original indoor purpose that it was conceived for and the limited outside work once you've learnt the necessary tricks.

    Adding Lidar as a supplementary or indeed replacement scanning technology would, even with cheap chips as mentioned above is going to push the tech up market and no doubt require a re-purposing of the existing software chain which I doubt will be trivial.

    I'm sure they're investigating the tech.

    Meanwhile are folk aware of this? LASiris VR about $13k USD iirc

    https://www.nctechimaging.com/lasiris/

    Not a competitor for Matterport yet but does create colourised cloud points and they are promoting an online service that they claim will create 3D models online with that data called OnestopVR but that isn't available yet.

  • That's a holy cow.  Thank you so much for sharing that information!

Please sign in to leave a comment.