4 ms·

X-Ray Vision for Robots: Seeing Through Walls with Only WiFi

34 points by pps 12y ago

ColinWright 12y ago
I can't see much difference between this and applying the standard CAT scan algorithms to signal strength received through structures. Can anyone tell me why this is particularly special? I agree it's clever, but I'm not seeing how it's much of a breakthrough.
- dm2 12y ago
  They say in the link that it's similar to how CT scans are done. I've never heard of anything similar to the dual robot method being done, it seems extremely impressive, is there any similar technology available? I'm guessing that the breakthrough is that it's using standard WiFi, the robots can be put into place remotely, and the size and shape of the room/object being scanned can be variable. As far as I know the WiFi imaging method has only been successful for about 5 years. The fact that it can go through concrete walls is also very impressive. It probably won't be useful for pinpointing people inside of a building (yet), but it would at least give soldiers, police, disaster teams, or construction teams a decent blueprint/layout of the building. Just another piece of information that could potentially help them do their job better. I think that they should attempt to create a more 3D map of the buildings using this technology, if possible. They could probably add several more sensors and inputs to create a better image. The processing time mentioned in the article seems very high, that might be one major limitation that better algorithms, methods, or hardware might be needed to fix.
  - ColinWright 12y ago
    > They say in the link that it's similar > to how CT scans are done. I didn't see that - thanks. The replies suggest that it really is "just" how CT scans are done, but with more uncertainty on the input, and therefore more processing to do to get the results. I've put this in my "should read more and work to understand" pile. Cheers.
- beambot 12y ago
  I cannot speak to their specific implementation.... but typically when one looks at classic tomography (eg. CT and MRI), there are tight tolerances or control over the relative pose of the transmitter and receiver. This is required for the deterministic radon transform equations. For a robot system (especially a dual robot system), you could imagine a lot of pose uncertainty (especially angular). Eg. if using laser rangefinders and AMCL localization, you'd typically find ~10cm and ~5-deg of non-gaussian pose uncertainty indoors. I could imagine this causing complications. I'd be very curious to see some form of probabilistic extensions to tomography (not that they don't exist; I've just not researched the topic).
  - ColinWright 12y ago
    Really useful reply - thank you.
endergen 12y ago
Am I missing something or is this research a waste of time? Impressive, but why waste time using Wifi when much better results are coming from structured light projector/scanner setups?
- spyder 12y ago
  Light projection doesn't go trough walls...
bane 12y ago
Really cool stuff. I wonder how well this would work with a pair of drones, or even a larger group of coordinated drones so you could "image" a multi-story building very quickly. Wi-fi radios are pretty small these days. Even relatively small quad copters could probably do this.
spyder 12y ago
Looks like the smaller but less accurate version of this: http://www.extremetech.com/extreme/100395-mit-can-now-see-through-concrete-walls http://www.extremetech.com/extreme/100395-mit-can-now-see-th...

34 points by pps 12y ago

I can't see much difference between this and applying the standard CAT scan algorithms to signal strength received through structures. Can anyone tell me why this is particularly special? I agree it's clever, but I'm not seeing how it's much of a breakthrough.

dm2 12y ago
They say in the link that it's similar to how CT scans are done. I've never heard of anything similar to the dual robot method being done, it seems extremely impressive, is there any similar technology available? I'm guessing that the breakthrough is that it's using standard WiFi, the robots can be put into place remotely, and the size and shape of the room/object being scanned can be variable. As far as I know the WiFi imaging method has only been successful for about 5 years. The fact that it can go through concrete walls is also very impressive. It probably won't be useful for pinpointing people inside of a building (yet), but it would at least give soldiers, police, disaster teams, or construction teams a decent blueprint/layout of the building. Just another piece of information that could potentially help them do their job better. I think that they should attempt to create a more 3D map of the buildings using this technology, if possible. They could probably add several more sensors and inputs to create a better image. The processing time mentioned in the article seems very high, that might be one major limitation that better algorithms, methods, or hardware might be needed to fix.
- ColinWright 12y ago
  > They say in the link that it's similar > to how CT scans are done. I didn't see that - thanks. The replies suggest that it really is "just" how CT scans are done, but with more uncertainty on the input, and therefore more processing to do to get the results. I've put this in my "should read more and work to understand" pile. Cheers.
beambot 12y ago
I cannot speak to their specific implementation.... but typically when one looks at classic tomography (eg. CT and MRI), there are tight tolerances or control over the relative pose of the transmitter and receiver. This is required for the deterministic radon transform equations. For a robot system (especially a dual robot system), you could imagine a lot of pose uncertainty (especially angular). Eg. if using laser rangefinders and AMCL localization, you'd typically find ~10cm and ~5-deg of non-gaussian pose uncertainty indoors. I could imagine this causing complications. I'd be very curious to see some form of probabilistic extensions to tomography (not that they don't exist; I've just not researched the topic).
- ColinWright 12y ago
  Really useful reply - thank you.

endergen 12y ago

Am I missing something or is this research a waste of time? Impressive, but why waste time using Wifi when much better results are coming from structured light projector/scanner setups?

spyder 12y ago
Light projection doesn't go trough walls...

bane 12y ago

Really cool stuff. I wonder how well this would work with a pair of drones, or even a larger group of coordinated drones so you could "image" a multi-story building very quickly. Wi-fi radios are pretty small these days. Even relatively small quad copters could probably do this.

spyder 12y ago

Looks like the smaller but less accurate version of this: http://www.extremetech.com/extreme/100395-mit-can-now-see-through-concrete-walls http://www.extremetech.com/extreme/100395-mit-can-now-see-th...