aliquot, post: 425055, member: 2486 wrote: What exactly do you mean by a 1cm "lock"?
I mean that the reported position holds true to within 1 centimeter.
As an example, if you send the robot on a mission and bring it back to the exact same place (at the exact same attitude), the position reported by k501g is within +- 1cm.
Does that make sense? Again, apologies if my language isn't precise to your field. Corrections appreciated 🙂
As others have mentioned, surveyors pay a lot for the convenience factor. Very few surveyors have the capability to gin up their own systems from OEM components. Even if they do, the time lost isn't worth it. We have a lot of different data collection and analysis pieces that we want to play together--GPS systems, total stations, AutoCAD, ArcGIS. All of these things can be had cheaply in OEM or generic forms--OEM boards, basic manual total stations, BricsCAD/IntelliCAD, and QGIS--but some of the connectivity breaks down.
Paying a lot for out-of-the-box systems with a smooth interface is kind of like getting older--it's better than the alternative.
FrozenNorth, post: 425058, member: 10219 wrote: kind of like getting older--it's better than the alternative.
FrozenNorth -- Are you my dad? Love it! Well said.
{lurker mode off}
I appreciate everyone who is able to put words together better than I, for your participation in this discussion.
{lurker mode on}
Roby,
Very interesting what you are doing.
The pictures of the trees in your back yard are nothing compared to where we desire to work. We want to be able to work deep in the woods with tree canopy 100 feet tall, all the way down to our knees, at the bottom of a ravine.
Multi-path mitigation is the one aspect where I would expect a big difference between the lower cost $500 boards,and the expensive guys. My Javad RTK receiver has 864 channels, and a half dozen parallel RTK engines. I am certain that there is a difference.
But...I do have a lot of grass to mow, and I would be thrilled if you could convert my mower into robotic mode. What are the average hardware costs to control a ZTR style (hydrostatic) mower? What "route planning" software do you use. I am also curious what your navigation system uses as its coordinate system. (UTM?)
John Evers, post: 425066, member: 467 wrote: What "route planning" software do you use. I am also curious what your navigation system uses as its coordinate system. (UTM?)
me too.
I worked with a company building robotic linear pivots.
An interesting project, way over my expertise.
I just provided RTK control using 'conventional' system for them to work with their gear.
I find this interesting, and whilst appreciate the research and development side I have always wondered as to the 'real' cost of RTK.
Thanks for posting Roby.
Roby, you may not want to use L2 boards that don't support L2C if you want them to work past 2020. The P(Y) signal at the L2 frequency may be retired sometime after Jan. 1, 2020. See http://www.insidegnss.com/node/3733 and http://gpsworld.com/to-l2c-or-not-to-l2c-that-is-the-operational-question/
Nate The Surveyor, post: 425045, member: 291 wrote: Research and development cost a lot of money. I bet a modern GPS system doesn't cost more than the grand in the hardware alone. All the rest goes for R&D. If you want cutting-edge equipment you're going to pay for R&D.
And marketing... how much does javad outspent their competition?
John Evers, post: 425066, member: 467 wrote: Roby,
Very interesting what you are doing.The pictures of the trees in your back yard are nothing compared to where we desire to work. We want to be able to work deep in the woods with tree canopy 100 feet tall, all the way down to our knees, at the bottom of a ravine.
Multi-path mitigation is the one aspect where I would expect a big difference between the lower cost $500 boards,and the expensive guys. My Javad RTK receiver has 864 channels, and a half dozen parallel RTK engines. I am certain that there is a difference.
But...I do have a lot of grass to mow, and I would be thrilled if you could convert my mower into robotic mode. What are the average hardware costs to control a ZTR style (hydrostatic) mower? What "route planning" software do you use. I am also curious what your navigation system uses as its coordinate system. (UTM?)
John, Those are *seriously* challenging conditions you describe! I would LOVE to see a no-holds-barred showdown by one of you guys who actually knows this stuff (not a marketing arm of any company, of course) staging the best products from Trimble/Novatel/ComNav/Javad against in each other in those conditions. Also, would be nice to see their prices side-by-side -- I have no idea what ComNav charges for comparable end-user products.
To your question about the mower. Here's a pic just snapped of an old prototype Scag Turf Tiger ("Rover 4") out in the shop:
To your question about coordinates -- the raw datum we process is WSG84 (I think that's it -- coordinate system acronyms never quite stick with me). The autopilot software (ArduPilot) cheats on a good bit of the math by approximating a flat earth and settling for centimeter-level precision on the navigation (i.e. we translate the initial coordinate to a coordinate plane centered at (0.00m, 0.00m, 0.00m) and do most of the maths in that plane -- this makes calculations a lot more tame (for instance, doing quaternion rotations to get from the antenna's position down to COG position -- sounds easy 'till you factor in tilt/roll/yaw)).
To your question about Route Planning -- most of that is Android with a little pure Java thrown in on the heavy lifting. Note, in the picture above you'll see a black tablet up in front of the driver (airplane cockpit style). So we have to roll our own Android app for the UI, and we've forked the Ardupilot code base using ArduPilot (seriously slick open source autopilot with Kalman filtering, etc) for the navigation engine on a variant of hardware called Pixhawk. If you look in the picture above you can just barely see a shiny box sitting just above the deck belts and below the seat -- that's a waterproof aluminum enclosure holding the autopilot hardware -- IMUs are much happier when close to the COG.
To your question about hardware costs. For Rover 4 above, let's approximate:
$12,000 -- Scag Turf Tiger 72"
$3,000 -- Sick TiM571 Lidar (Obstacle avoidance -- NOT pictured -- mounts on big bar across front)
$890 -- ComNav k501g RTK GPS/GLONASS L1/L2 (@10hz)
$150 -- L1/L2 GPS/GLONASS Antenna
$1,500 -- ALL other electronic components (heavy duty Screen, Servos, Radios, companion ARM computers, etc)
If you don't have access to a corrections source, it's going to cost another $1000 to roll a good L1/L2 GPS/GLONASS correction station with receiver/antenna/cables/enclosure/screen. We have access to both public corrections (Alabama CORS) and we have our own base stations -- of course we use our own 🙂
If you think about the price list above -- 5 years ago, instead of $2000 for ALL GPS components, you would have been lucky to get them for $10,000 (wild guess). 5 years ago I'll guess $7,000 for the Lidar. Lidar is amazing stuff (not quite as amazing as RTK GNSS imo, but close) and Velodyne/Quanergy/et al are in a wild race to get it really cheap -- expectation (admittedly fickle) is an order of magnitude cost reduction on the horizon and mass production of solid state units.
So, back to your grass mowing needs, just hang on a few years my friend 🙂
--Roby
I hope the GPS on board the mower can detect that metal baseball bat the kids forgot to bring inside last night.... or the neighborhood stray cat that decided to take a snooze on your overgrown lawn during mow time
Sent from my SM-G955U using Tapatalk
Rich., post: 425089, member: 10450 wrote: I hope the GPS on board the mower can detect that metal baseball bat the kids forgot to bring inside last night.... or the neighborhood stray cat that decided to take a snooze on your overgrown lawn during mow time
Sent from my SM-G955U using Tapatalk
That's LIDAR, my friend.
Roby_, post: 425090, member: 12678 wrote: That's LIDAR, my friend.
Speaking of 'autopilot' and object detection, I took a test drive in a Tesla Model X yesterday and tried out the autopilot feature... it was incredible, although I was a bit too comfortable with my hands not on the wheel while on the highway.
Sent from my SM-G955U using Tapatalk
Roby_, post: 425009, member: 12678 wrote: Jim -- Can you advise if there's a reason why you guys pay that much? Please don't read that as being sarcastic -- being totally sincere -- I realize I'm playing a game here where you are the expert -- it's just hard for me (as a programmer/robotic builder) to understand what benefit you guys get by paying so much for equipment.
Build something durable and reliable and we will probably buy. I'm a surveyor with very limited knowledge of what it takes to put a system together. It's easier to make money at something I know then save money on something I don't. Thanks for sharing this interesting article.
Rich., post: 425091, member: 10450 wrote: Speaking of 'autopilot' and object detection, I took a test drive in a Tesla Model X yesterday and tried out the autopilot feature... it was incredible, although I was a bit too comfortable with my hands not on the wheel while on the highway.
Sent from my SM-G955U using Tapatalk
Rich -- My envy-dar is pegging with your Tesla test drive. Hey, as soon as I shot my terse response above about LIDAR I regretted it. It simply was not fair to your post. Specifically, we've probably devoted more time (and sleepless nights) to obstacle detection and avoidance than any other single aspect of robotics. It's a huge problem to solve: LIDAR / RADAR / VISION (Mobileye/OpenCV/etc) / even SONAR all have a place in that discussion and it's being aggressively worked on as we speak. Tesla is pushing the boundaries, but they've had some high-profile autopilot fails. Obstacles (especially the irreplaceable ones) are the things that have to keep robot builders humble.
Roby,
You can add Topcon to your list.
I'm sure you'll get a taker on the no holds barred head to head GNSS challenge, but it won't come without a advertising. This board is virtually run by the Javad team, so be prepared to be unindated by their superior products.
A Harris, post: 424986, member: 81 wrote: Howdy Roby,
Glad to hear a new angle on RTK.When Kris sees this he will want one for his Bobcat..........
Right now, I'm working on getting the snakes out of it so I can drive it.......
[USER=12678]@Roby_[/USER]
Your writing style is great. I wish more technical writers would learn your ability to make articles interesting. 🙂
Oh and you forgot France (upcoming)..."considering the events of the last few years (Greece/Brexit), itÛªs success may not be a foregone conclusion." 😉
FL/GA PLS., post: 425115, member: 379 wrote: [USER=12678]@Roby_[/USER]
Your writing style is great. I wish more technical writers would learn your ability to make articles interesting. 🙂
Oh and you forgot France (upcoming)..."considering the events of the last few years (Greece/Brexit), itÛªs success may not be a foregone conclusion." 😉
Wow, thanks bro -- you made my morning. Hey, I'm ridiculously indebted to you guys for nearly every piece of RTK knowledge. I had hoped you guys would weigh in a little on that article -- but all the info you all have pointed out (and gaps in my thinking, especially) has been awesome -- much appreciated!
Roby_, post: 425093, member: 12678 wrote: Rich -- My envy-dar is pegging with your Tesla test drive. Hey, as soon as I shot my terse response above about LIDAR I regretted it. It simply was not fair to your post. Specifically, we've probably devoted more time (and sleepless nights) to obstacle detection and avoidance than any other single aspect of robotics. It's a huge problem to solve: LIDAR / RADAR / VISION (Mobileye/OpenCV/etc) / even SONAR all have a place in that discussion and it's being aggressively worked on as we speak. Tesla is pushing the boundaries, but they've had some high-profile autopilot fails. Obstacles (especially the irreplaceable ones) are the things that have to keep robot builders humble.
Yep, that roomba is great too until it goes at a pile of dog crap and swirls it across the floor.
Also, knowing where Kris lives, he literally means that he is removing reptiles from his bobcat. That wasn't an East Texas euphemism for debugging the equipment.
Wasn't there a guy in Eastern Europe on here who was building his own GPS receivers? Can't remember the details but he seemed to be going for OEM boards and casting/molding his own antennas.
