That is interesting. About 20 years ago, I was very involved in the GPS newsgroups that had both GPS engineers and scientists as well as consumers discussing the emerging technologies. We talked about the theory of building indoor GPS transmitters to be able to locate objects/people inside buildings. The problem at the time was that the processors on the receivers were not fast enough to measure the time difference between their internal clock and the time stamp of the received signal that hadn't traveled 20,200 km. The capabilities of the military units that used the phase differences between two different frequencies on the same bird to get higher accuracy was still classified, so I am sure some of the engineers knew that this kind of thing could be done, but had to bite their tongues. Of course at the time, we were talking about dedicated transmitters. This idea of using existing transmitters and receivers, and the much much faster processors available today was something we would not have even dreamed about back then.
Can I use this technology to successfully navigate my environment, giving me accurate indoor gps information that would allow me to get through a typical doorway without bashing into the wall instead? The answer is no.
Don't get me wrong, I'm impressed, but from what I understand about it Chronos has minimal and limited uses. I won't be holding my servos (breath) for this one. It's brilliant and innovative, but doesn't seem to be a game changer for indoor robots that arn't drones or drones that need to exit or enter different rooms. And if one of the wifi points went down for any reason, what have you got then?
Any flying robot that has up to 98cm error in location judgements would not be permitted near my human, that's for sure. Perhaps I have missed something or mis-understood the paper in some way and if so, I would love to hear about it.
@Girard,
You are correct but short sighted IMHO. Autonomous robots shouldn't depend on only one sensor for info. Without a map and ping sensors or camera of some sort, you wouldn't even know a door was there.
The use would be that when used with SLAM, you could have another method of knowing where you are. It could even be used as an initial indication and then use SLAM to get a better indication. It could be used for the robot to know where you are in the house based on where your cell phone is.
This technology isn't going to give you a map of your surroundings just like a GPS receiver doesn't give you the map. Software gives you the map, GPS just gives you a point in a coordinate based system [edit] and software makes adjustments to this coordinate to draw the image of your car on the image of the road.[/edit]
[edit]
Without multiple methods to verify sensors are accurately working, any failure of a device can have bad impacts. A WiFi access point not being available is the equivalent of a ping sensors or camera going out. Other means of validation as to your location are necessary for any device failure.
[/edit]
'Our goal is to design a system that enables a single WiFinode (e.g.,anaccesspoint) to localize another, with-out support from additional infrastructure. Further, we would like a design that works on commodity WiFiNICs and does not require any additional sensors (cameras,accelerometers,etc.).' Chronos.
'Autonomous robots shouldn't depend on only one sensor for info. Without a map and ping sensors or camera of some sort, you wouldn't even know a door was there. The use would be that when used with SLAM, you could have another method of knowing where you are. It could even be used as an initial indication and then use SLAM to get a better indication.' CochranRobotics.
And thus Chronos is not in and of itself a game changer, but tech that's comparable to what's already available.
Perhaps a key word here is 'infrastructure'. Your suggested SLAM would be considered infrastructure.
No, SLAM is software, not infrastructure. It uses sensors, not infrastructure.
Sure, other things are available. RFID chips being carried around for example, instead of cell phones on WIFI network.
Infrastructure would be RFID radios or network, servers or other hardware devices. SLAM, like GPS uses software to calculate locations based on "sensors".
In any event, it is another method of giving localization information, which can the be validates through other means.
This isn't developed specifically for robotics. Like everything else, you adopt multiple technologies to help your robot. Ping sensors, LIDAR, GPS, sonar sensors, compasses and so many other things (DC motors, pots, radios, cameras) were not specifically developed for Robotics. Everything has been adapted to work with robotics. It is a technology that can be used to help your robot determine things (location). Even if this is within 2 feet, it can still be helpful. Disregarding new technology without trying it out is not wise.
I don't think that anyone thinks that you would just drop in this one device and your robot would suddenly be all knowing. It is another bit of information to help make your robot smarter. It is the combination of all of your sensors that makes the robot smart. Depending on one sensor makes for a "dumb" robot irregardless of what it is.
[edit] Edited for clarification of a term used... dumb = not able to validate the information that it is using to make decisions based on limited information. Also edited to fix things cause by typing from a cell phone. [/edit]
That is interesting. About 20 years ago, I was very involved in the GPS newsgroups that had both GPS engineers and scientists as well as consumers discussing the emerging technologies. We talked about the theory of building indoor GPS transmitters to be able to locate objects/people inside buildings. The problem at the time was that the processors on the receivers were not fast enough to measure the time difference between their internal clock and the time stamp of the received signal that hadn't traveled 20,200 km. The capabilities of the military units that used the phase differences between two different frequencies on the same bird to get higher accuracy was still classified, so I am sure some of the engineers knew that this kind of thing could be done, but had to bite their tongues. Of course at the time, we were talking about dedicated transmitters. This idea of using existing transmitters and receivers, and the much much faster processors available today was something we would not have even dreamed about back then.
Alan
yep, being able to measure time in nanoseconds is important here. The processors of today can handle it much better.
I am looking forward to this technology becoming publicly available. It is a game changer in robotic applications. I just hope it doesn't flop.
I have heard that Cisco is working on this already. May have to switch back to Cisco if they can make this happen.
This technology is interesting and very clever but the articles title is typical click-bait. An 'accurate' indoor GPS would not have a median error of 65cm in line-of-sight and a 98cm median error in non-line-of-sight. Source: https://www.usenix.org/system/files/conference/nsdi16/nsdi16-paper-vasisht.pdf
Can I use this technology to successfully navigate my environment, giving me accurate indoor gps information that would allow me to get through a typical doorway without bashing into the wall instead? The answer is no.
Don't get me wrong, I'm impressed, but from what I understand about it Chronos has minimal and limited uses. I won't be holding my servos (breath) for this one. It's brilliant and innovative, but doesn't seem to be a game changer for indoor robots that arn't drones or drones that need to exit or enter different rooms. And if one of the wifi points went down for any reason, what have you got then?
Any flying robot that has up to 98cm error in location judgements would not be permitted near my human, that's for sure. Perhaps I have missed something or mis-understood the paper in some way and if so, I would love to hear about it.
@Girard, You are correct but short sighted IMHO. Autonomous robots shouldn't depend on only one sensor for info. Without a map and ping sensors or camera of some sort, you wouldn't even know a door was there.
The use would be that when used with SLAM, you could have another method of knowing where you are. It could even be used as an initial indication and then use SLAM to get a better indication. It could be used for the robot to know where you are in the house based on where your cell phone is.
This technology isn't going to give you a map of your surroundings just like a GPS receiver doesn't give you the map. Software gives you the map, GPS just gives you a point in a coordinate based system [edit] and software makes adjustments to this coordinate to draw the image of your car on the image of the road.[/edit]
[edit] Without multiple methods to verify sensors are accurately working, any failure of a device can have bad impacts. A WiFi access point not being available is the equivalent of a ping sensors or camera going out. Other means of validation as to your location are necessary for any device failure. [/edit]
'Our goal is to design a system that enables a single WiFinode (e.g.,anaccesspoint) to localize another, with-out support from additional infrastructure. Further, we would like a design that works on commodity WiFiNICs and does not require any additional sensors (cameras,accelerometers,etc.).' Chronos.
'Autonomous robots shouldn't depend on only one sensor for info. Without a map and ping sensors or camera of some sort, you wouldn't even know a door was there. The use would be that when used with SLAM, you could have another method of knowing where you are. It could even be used as an initial indication and then use SLAM to get a better indication.' CochranRobotics.
And thus Chronos is not in and of itself a game changer, but tech that's comparable to what's already available.
Perhaps a key word here is 'infrastructure'. Your suggested SLAM would be considered infrastructure.
Giskard.
No, SLAM is software, not infrastructure. It uses sensors, not infrastructure.
Sure, other things are available. RFID chips being carried around for example, instead of cell phones on WIFI network.
Infrastructure would be RFID radios or network, servers or other hardware devices. SLAM, like GPS uses software to calculate locations based on "sensors".
In any event, it is another method of giving localization information, which can the be validates through other means.
This isn't developed specifically for robotics. Like everything else, you adopt multiple technologies to help your robot. Ping sensors, LIDAR, GPS, sonar sensors, compasses and so many other things (DC motors, pots, radios, cameras) were not specifically developed for Robotics. Everything has been adapted to work with robotics. It is a technology that can be used to help your robot determine things (location). Even if this is within 2 feet, it can still be helpful. Disregarding new technology without trying it out is not wise.
I don't think that anyone thinks that you would just drop in this one device and your robot would suddenly be all knowing. It is another bit of information to help make your robot smarter. It is the combination of all of your sensors that makes the robot smart. Depending on one sensor makes for a "dumb" robot irregardless of what it is.
[edit] Edited for clarification of a term used... dumb = not able to validate the information that it is using to make decisions based on limited information. Also edited to fix things cause by typing from a cell phone. [/edit]
Thank you for taking the time to explain. I appreciate it.