Hi Curtis, Curtis Olson wrote: > Hi Tiago, > > Just a real quick reply because no one else ahs ... > > On Mon, May 14, 2012 at 4:59 PM, Tiago Gusmão wrote: > > Hello, > > It's been a long time since i've posted here! > > I'm currently employed in a small company here in Portugal > (http://www.albatroz-eng.com/). Our main products in a system that > inspects powerlines from helicopters. > This year we are going to a big exhibition and i've suggested that it > would be interesting to have a simulator (only because we would have a > hard time flying the helicopter inside the building ;) ). > > Our solution is basically comprised of a LiDAR, an INS/GPS unit and > camera(s). > > Simulating INS/GPS is easy because all the data is in the FG tree AFAIK. > > > Yes, FlightGear provides "exact" roll, pitch, yaw (true) as well as lon, > lat, and altitude. > > > Camera(s) can be linked instances of FG with the views properly set > (perhaps captured with image server). > > > I've done some work with setting up a model of a pan/tilt camera system > that can point at a specific wgs84 point or along a specific NED vector > (i.e. nadir, or exactly at my shadow, etc.) This was [unfortunately] > for a paid consulting project so that code doesn't live in the > FlightGear tree. However, it's really easy to configure a view that > stays locked on a specific lon/lat and I hacked a small bit of nasal to > copy the point you click on over into the view target variables so you > can click any where in the scene and the pan/tilt camera will hold > center on that exact location. FlightGear offers a lot of flexibility > and comparability in this arena.
I was only considering cameras that are fixed to the aircraft but we do integrate video feeds from gimbal cameras so this might be a nice idea and, as you say, probably not that hard to do. > > > LiDAR is the only part the gets tricky. Our LiDAR is placed pointing to > the side and down of the aircraft. It's a 2D LiDAR that captures > "slices" of the powerline and as the helicopter flies forward we get all > the line (like this: > http://www.albatroz-eng.com/solutions/plmi/lidar_ohl_example.PNG). > My ideia is to have an instance of FG with the camera pointed as if it > was the LiDAR and then capture a column of the Z-buffer. > > From what i understand FG has depth partitioning. I'm guessing we don't > want this active for this application. Is this correct? Is there any > "switch" for this or will i need to dive into the code? The LiDAR has > about 150m of useful range. > > Is there any "recommended" method of getting the values from the > z-buffer? From what i hear some methods have very low performance. Has > anyone tinkered with this? > > > FlightGear has the ability to intersect a vector with the terrain and > return the exact altitude of the "hit". This is essentially what we are > doing when we compute the location of the point we click on with the > mouse. However, it may not scale well to the number of intersection > tests you would need to do (per frame) for a scanning lidar system. > > Perhaps there could be a z-buffer trick you could use, but I don't know? > Maybe Fred (who's working on shadows and lighting effects) might be the > best one to comment on that since he has his head in that type of code > right now ... (?) The concept of using the z-buffer seems easy: get the pixel data with z ,apply the inverted projection matrix and from those coordinates get a distance and angle and it's done. It's the details that concern me :) > > The intersection test code does cache small bits of scenery that was > most recently used, so if you do scan across a line of terrain quickly > you might get a bunch of cache hits and have a chance at making this > work, but you'd definitely need to test it out before putting your eggs > in that basket. > > I think the problem will be doing this at a resolution that could pick > out individual power lines. I did consider that we can do intersection. We take ~400 points at 25fps in the LiDAR. I think we would be lucky to get any cables in the simulator with those sampling intervals unless we would actually model them as thick cylinders (which would impact performance a lot if we make cables with lots of segments). > > It would be possible to create models of trees and shrubs and connected > power lines and populate an area of the scenery with your custom models > -- it would be some effort, but if you can make (or locate) quality > models, the result would be quite nice. What is the state of the art for populating a map with models? I seem to remember being able to point-and-click on the terrain to "drop" a model but is there any way to make the changes permanent? > > Here's another random idea -- what about "faking" the LiDAR data or at > least some of it? I'm not sure you would have the performance to > physically model everything the way it is in real life, but if you have > the aircraft location and orientation and an offline DB of your power > lines, perhaps you could compute "fake" LiDAR data from that using > special purpose code, and that could feed your LiDAR display? It's an interesting concept, we could perhaps fake the line itself but the surroundings would be more tricky as we can't be sure the simulated helicopter would fly in the exact same path. It has occurred to me that maybe we can take LiDAR data and process it to get some sort of models. But perhaps that would be an R&D project by itself! > > I'm sure there are several ways you could attack this and it may just > mean picking the best likely approach, diving in, and then seeing how > far you can get with it? Agreed. Unfortunately time is a bit short so i hope to pick the right approaches the first time. I am confident that this will work sooner or later even if we have to throw some powerful HW for it. Thank you for your help! Best regards, Tiago > > Best regards, > > Curt. > > > For the scenery we would need a powerline with believable trees, shrubs > and houses below and nearby (and they obviously can't be done with some > trick that results in no z-buffer entry). The scenery doesn't need to > match real world locations of powerlines. > > Do we have any part of the scenery that is suited for this? > Last time i've seen we have towers but no cables between them. Is this > still true? How should we go about solving this? > I'm guessing i could make the powerline as an extremely big model but if > it can be done with code i think we would be better served (and it would > be a nice addition to FG) > If we could crash into the powerlines it would be a good feature but > it's not required > > As for helicopters, is anyone working on a EC350, EC120 or Bell 206? > These are the types we have used and so would prefer to have in the sim > but it's just a minor detail. > > > > Thanks! > > Cheers, > Tiago > > > > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. > Discussions > will include endpoint security, mobile security and the latest in > malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Flightgear-devel mailing list > Flightgear-devel@lists.sourceforge.net > <mailto:Flightgear-devel@lists.sourceforge.net> > https://lists.sourceforge.net/lists/listinfo/flightgear-devel > > > > > -- > Curtis Olson: > http://www.atiak.com - http://aem.umn.edu/~uav/ > http://www.flightgear.org - http://gallinazo.flightgear.org > > > ------------------------------------------------------------------------ > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > > > ------------------------------------------------------------------------ > > _______________________________________________ > Flightgear-devel mailing list > Flightgear-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/flightgear-devel ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
