nivek@rover.ri.cmu.edu.UUCP (05/13/88)
Cameras and telemetry with mobile robots- I've used a number of CCD cameras because they have other advantages over the tube cameras. Environmental considerations, size, weight, lack of blooming in the image (w/ IR cut filters) and more recently resolutions that approach broadcast quality. As far as the transfer method (frame or line), it's been our experience that to use standard video outweighs digitizing artifacts like aliasing. You won't be tied into specific hardware or interfaces, or code for an off-the-shelf digitizer. Buy instead of make is a much more productive philosphy in mobile robots. You can spend too much time making your own motors, or hardware, or interfaces, etc etc etc. It's not usually worth it unless the goal is to do research or work in motors, interfaces, etc. There are now small CCD cameras with 512x512 resolution or greater that are very small (Sony, the Pulnix cameras are mostly modified Sonys) With a normal digitizer of that resolution and image reduction to 256x256 you should see few if any artifacts from digitizing. In one test set-up on a robot using scan-line stereo techniques we have found a slight jitter in the constructed 3D image that may be due to aliasing in the digitizer, but this seems to be an extreme case and I haven't noticed this in other work here. Transmission- There are some illegal transmitters for video out there mostly sold through a couple of surplus places. We obtained an experimental FCC license that encompasses several microwave, radio and televsion frequencies. We have used our low power UHF transmitters (channels 24 and 46) for transmitting stereo! Microwave is very nice, works well indoors, through (sheetrock) walls, but is mostly line-of-sight. We proposed a large-scale telemetry system for a proposed rough-terrain forklift that encompassed large bandwidth communication between the RT-FL and a base station. The system was to be auto-tracking as well. We didn't get that particular proposal though and didn't get a chance to build it...yet... The microwave system we have has a small omni-directional transmitter and needs 24VDC and a video input. Very simple to use and pretty small. Not sure about FM TV gear for standard video you need a lot of bandwidth and I'm not sure FM bands will do it. A big decision is where in the processing chain to split the system. If you do all computing off-board that means high-bandwidth telemetry but you are unlimited in your off-board possibilities. You can do some preprocessing onboard and send less information (road edges, center of mass, or whatever symbolic information you're trying to extract from images) by radio modem or lower bandwidth transmission. Stabilization- There are a few systems. The academy award winning Steadi-Cam is a lease only system and you get a qualified 'operator' with it. Not very tractable for autonomous mobile robots! (Garrett Brown, the inventor, also has this neat sky-cam controlled by several cables above a stadium.) Gyro-zoom and some other systems use optics and mirrors. There are a number of military devices developed for helicoptor people to provide a rock steady image in severe vibration. These are built into cameras and binoculars. It turns out though that image stability hasn't been a problem though. Most mobile robots move pretty slow and suspension in tires, frame etc do a remarkable job of smoothing motions. On two of our systems, Terregator and the Navigation Laboratory (NavLab) we had shear loaded shock mountings for sensors. These worked very well. Basically it is a platform for sensors that is mounted to a frame but isolated by shear loading elastomer mounts. Sending data- Up to 9600 baud or so you can use radio modems, a number of which are commercially available. We got a cellular phone for communication to researchers at the vehicles which helped a great deal, and it is possible to get cellular modems now (up to 2400 baud, I think) One of our laser range scanners though, put out high speed data that was too much for the commercial market devices. I wrote a proposal to the state for high-tech development and we worked with a local transmitter company and developed a device to send data on the audio subcarrier of one of our UHF licenses. It worked well, but we built the NavLab since then which is a large computer controlled van that houses locomotion, computers, power and researchers. So it didn't need off-board telemetry although we could show pretty graphics during the image and range processing to off-board audiences using these links. If you need info on the local company let me know. After more than a dozen or so complete robot systems I find the tendency is to emplace more and more computing onboard. Telemetry is costly not only in money but size, weight and power might be put towards computers instead. Telemetry is also another component in your system, one that requires a lot of overhead. nivek Aka : Kevin Dowling Bell: (412) 268-8830 Arpa: nivek@rover.ri.cmu.edu Mail: Robotics Institute Carnegie Mellon University Pgh, PA 15213-3890
Pretty-Boy@cup.portal.com (05/20/88)
The question at hand, is "How do I get video from here to there?"
As a licensed Amateur (Tech class or higher) you have priviledges for
Full TV transmission! Color, sound, whatever you want ... on freq's
up around 400 MHz. Also, you can put your telemetry down on 220 Mhz
using stock Packet gear (There's a newsgroup for packet-radio)
There is a company out there (Not on my desk rite now) that sells a small
10 watt or so TV xmitter (Crystal controlled) in the 400 Mhz band, for about
$80 or so ... (Don't quote me) ..... E-Mail if interested ...
Good luck ....
pretty-boy@cup.portal.com