Loading ...
Sorry, an error occurred while loading the content.

Re: [SeattleRobotics] Neato wifi

Expand Messages
  • Bruce J Weimer MD
    Dave, Thanks for the information! I don t really need a processor on the Neato since I plan to do all the processing on the desktop. I just need to be able to
    Message 1 of 11 , Jul 1 12:28 PM
    • 0 Attachment
      Dave,

      Thanks for the information!

      I don't really need a processor on the Neato since I plan to do all the
      processing on the desktop. I just need to be able to link the desktop to
      the Neato somehow (somehow other than a really long USB cable). So the
      question is: Is there a Linux compatible USB wifi dongle that I could use?
      Has anyone done this? Any information or better ideas would be appreciated!

      Dr. Bruce.
    • Dave Curtis
      ... If you go with a Chumby, any RT73-based dongle should be plug-and-play. If you go to the Adafruit wiki site, there are a couple of tutorials on using a
      Message 2 of 11 , Jul 1 6:09 PM
      • 0 Attachment
        On Jul 1, 2011, at 12:28 PM, Bruce J Weimer MD wrote:

        > Dave,
        >
        > Thanks for the information!
        >
        > I don't really need a processor on the Neato since I plan to do all the
        > processing on the desktop. I just need to be able to link the desktop to
        > the Neato somehow (somehow other than a really long USB cable). So the
        > question is: Is there a Linux compatible USB wifi dongle that I could use?

        If you go with a Chumby, any RT73-based dongle should be plug-and-play. If you go to the Adafruit wiki site, there are a couple of tutorials on using a Chumby or Chumby Hacker Board with WiFi. One is by me, on how to set up the Chumby to serve as an AP but without any back-haul, which allows the robot to serve as the AP for it's own WiFi network, which is very convenient since that way you don't need to cart along an AP and have another piece of equipment that can fail or be forgotten or have flat batteries or...

        The Chumby One came with an RT73 WiFi stick in it, if you can still find one of those to hack. Sparkfun had refurb Chumby RT73 WiFi dongles (no case, just naked PCB) for under $10 for a while, check their stock situation, they may be out. I printed a very nice case for the Sparkfun dongle on my Makerbot Cupcake, but haven't gotten around to uploading the design to Thingiverse.

        Also, google with terms something like "bob smith chumby robot" or such... Bob was the original pioneer of using Chumby's on robots. He did a nice paper on basic chumby set up that is directly applicable to robot hackers. That white paper is the root reference to most chumby-based robot hacking. Ironically, I based my Chumby access point write up on a blog posting by another guy that based his post on Bob's paper -- a few days later I saw Bob at an HBRC meeting and he was using my wiki article to bring up a new Chumby -- closing the loop :)

        Hash79's blog, "Random Workshop" if I recall correctly, has a posting about doing exactly what you want -- using the Chumby as a wireless serial cable to talk to an XV-11. (Hash is the guy that won the XV-11 Lidar hacking prize.) He found a nice little telnet redirect package that works great for wireless serial applications. So if you can get an RT73-based dongle, which is about the only one that reliably does AP mode on the Chumby, then set it up for AP mode, and add Hash's telnet redirector, you should be all set. Any WiFi device can join the robot's network, telnet in, and talk to the XV-11 directly.

        Good luck!

        -dave



        > Has anyone done this? Any information or better ideas would be appreciated!
        >
        > Dr. Bruce.
        >
        >
        >
        > ------------------------------------
        >
        > Visit the SRS Website at http://www.seattlerobotics.orgYahoo! Groups Links
        >
        >
        >
      • Weston Turner
        If you are not embedding significant compute power in the robot, why would you want to use a heavyweight TCPIP stack and wifi? Why not use a lightweight
        Message 3 of 11 , Jul 1 6:10 PM
        • 0 Attachment
          If you are not embedding significant compute power in the robot, why would you want to use a heavyweight TCPIP stack and wifi? Why not use a lightweight wireless protocol like Zigbee or Bluetooth, which essentially create a wireless RS232 connection between the remote computer and the robot? You can get 100m range out of Bluetooth, and over 1.5km range with Zigbee. Furthermore, the variety of network topologies one can implement using Zigbee makes it a very appealing for developing swarm robots and other distributed robotics applications.

          Weston

          On Fri, Jul 1, 2011 at 12:28 PM, Bruce J Weimer MD <bjweimer@...> wrote:
          Dave,

          Thanks for the information!

          I don't really need a processor on the Neato since I plan to do all the
          processing on the desktop.  I just need to be able to link the desktop to
          the Neato somehow (somehow other than a really long USB cable).  So the
          question is: Is there a Linux compatible USB wifi dongle that I could use?
          Has anyone done this? Any information or better ideas would be appreciated!

          Dr. Bruce.



          ------------------------------------

          Visit the SRS Website at http://www.seattlerobotics.orgYahoo! Groups Links

          <*> To visit your group on the web, go to:
             http://groups.yahoo.com/group/SeattleRobotics/

          <*> Your email settings:
             Individual Email | Traditional

          <*> To change settings online go to:
             http://groups.yahoo.com/group/SeattleRobotics/join
             (Yahoo! ID required)

          <*> To change settings via email:
             SeattleRobotics-digest@yahoogroups.com
             SeattleRobotics-fullfeatured@yahoogroups.com

          <*> To unsubscribe from this group, send an email to:
             SeattleRobotics-unsubscribe@yahoogroups.com

          <*> Your use of Yahoo! Groups is subject to:
             http://docs.yahoo.com/info/terms/


        • Nathan Monson
          ... Zigbee is too slow, since the XV-11 can generate over 30KB of sensor data per second. There s a lot more going on than in a Roomba. I d be interested to
          Message 4 of 11 , Jul 1 11:55 PM
          • 0 Attachment
            On Fri, Jul 1, 2011 at 6:10 PM, Weston Turner <wstnturner@...> wrote:
            If you are not embedding significant compute power in the robot, why would you want to use a heavyweight TCPIP stack and wifi? Why not use a lightweight wireless protocol like Zigbee or Bluetooth

            Zigbee is too slow, since the XV-11 can generate over 30KB of sensor data per second.  There's a lot more going on than in a Roomba.

            I'd be interested to hear about your experiences using Bluetooth over 100m. Which product did you use?  I've attempted to use Bluetooth in a couple of robots and just can't get the range or reliability needed.  Indoors, it is prone to bursts of lag as the robot moves around, which just sucks with a mobile robot that needs good responses.  I've used USB Bluetooth dongles and standalone UART modules.

            For me, only Wifi has been sufficiently reliable/low-latency/functional-in-homes.  And of course it's not perfect either, but it sure beats Bluetooth!

            Nathan
          • Weston Turner
            The XBee Pro 50mW RPSMA - Series 2 supports 30 kilobytes/second. It has a max data rate of 250kb/s. http://www.sparkfun.com/products/10419 And: (250*1024) /
            Message 5 of 11 , Jul 2 3:16 PM
            • 0 Attachment
              The XBee Pro 50mW RPSMA - Series 2 supports 30 kilobytes/second. It has a max data rate of 250kb/s. http://www.sparkfun.com/products/10419

              And:
              (250*1024) / (30*1024*8) = 256000 / 245760. The spec supports the requirement, but maybe that is not enough headroom for the application. 802.11b gives you about 45 times more bandwidth, more flexibility if you want to extend the project by embedding more controllers, sensors, etc. 

              I don't have personal experience with long range Bluetooth communication. Only read specs on modules such as:

              SparkFun's Bluetooth Modem - BlueSMiRF Gold:
              The data-sheet claims it "delivers up to 3 Mbps data rate for distances to 100M."

              I've used short range Bluetooth communication for remote control and robotic telemetry applications without latency issues. In fact I was surprised at the responsiveness of the system (tens of milliseconds), but maybe that is not fast enough for this application.  

              Using Wifi will give you much more bandwidth and flexibility but at a cost of a greater integration effort. Form a software engineering perspective, you would go from interfacing with the simple RX/TX pipe of Bluetooth SPP to developing an interface to the command set of whatever Wifi module you chose to use. In other words, chose between something I would spend my free time playing with to something I would get paid to work on for eight ours a day five days a week.

              Also, I have doubts about the actual bandwidth needs of the application. Surely some pre-processing could be applied to the data stream to make it sparser. Often using LIDAR is like sipping from a fire-hose, it provides you with an overwhelming amount of data, while only a small subset of it is actually useful for navigation and object avoidance.

              Best Regards,
              Weston

                
              On Fri, Jul 1, 2011 at 11:55 PM, Nathan Monson <nmonson@...> wrote:
               

              On Fri, Jul 1, 2011 at 6:10 PM, Weston Turner <wstnturner@...> wrote:
              If you are not embedding significant compute power in the robot, why would you want to use a heavyweight TCPIP stack and wifi? Why not use a lightweight wireless protocol like Zigbee or Bluetooth

              Zigbee is too slow, since the XV-11 can generate over 30KB of sensor data per second.  There's a lot more going on than in a Roomba.

              I'd be interested to hear about your experiences using Bluetooth over 100m. Which product did you use?  I've attempted to use Bluetooth in a couple of robots and just can't get the range or reliability needed.  Indoors, it is prone to bursts of lag as the robot moves around, which just sucks with a mobile robot that needs good responses.  I've used USB Bluetooth dongles and standalone UART modules.

              For me, only Wifi has been sufficiently reliable/low-latency/functional-in-homes.  And of course it's not perfect either, but it sure beats Bluetooth!

              Nathan


            • Nathan Monson
              ... This is the very module I used. Maybe it was something about the environments I tested in. Both at the apartment and in the office, performance was just
              Message 6 of 11 , Jul 2 3:33 PM
              • 0 Attachment
                On Sat, Jul 2, 2011 at 3:16 PM, Weston Turner <wstnturner@...> wrote:
                The data-sheet claims it "delivers up to 3 Mbps data rate for distances to 100M."

                This is the very module I used.  Maybe it was something about the environments I tested in.  Both at the apartment and in the office, performance was just peachy until line of sight was lost.  There are plenty of other things chattering at 2.4GHz in both places, and metal cubicle frames can't be too good.
                 
                Using Wifi will give you much more bandwidth and flexibility but at a cost of a greater integration effort.

                I'm not so sure about this in the case of the XV-11.  One hacker found out that the XV-11 does have a UART, but it runs at a slow speed.  The 'high speed' sensor data can't exit this port.  The USB port is the only way to access the laser.

                Since you must have a USB host stack to get the laser sensor, the integration is already beyond just gluing things together.

                But a Chumby with WIFI is almost the perfect fit.  The Chumby recognizes the XV-11 as soon as you plug it in, and a couple of lines of shell scripting connects it over WIFI.  That's some nice integration!
                 
                Also, I have doubts about the actual bandwidth needs of the application. Surely some pre-processing could be applied to the data stream to make it sparser.

                This is very true, and a good idea no matter what protocol is used.  The XV-11 seems to use verbose ASCII to send data, and quite a bit of the data is of dubious or unknown value, such as error codes and laser intensity values.

                Of course, the Chumby makes that pretty easy too.

                Cell phones are another interesting choice.  I met somebody who hacked his Droid (software) to be a USB host, so any Chumby code would work equally well on that platform.

                Nathan
              • Weston Turner
                Yes, using a smart-phone running Droid to interface with the robot would be very exiting. The Chumby just seems a little clunky, based on the blog and
                Message 7 of 11 , Jul 2 4:00 PM
                • 0 Attachment
                  Yes, using a smart-phone running Droid to interface with the robot would be very exiting. The Chumby just seems a little clunky, based on the blog and complaints about it being power hungry, needing active cooling, etc..

                  I was under the impression that the USB and serial interfaces provided the same level of access to the VX-11. Since they don't, and one cannot retrieve sensor data via the serial interface, that pretty much requires using an intermediate controller with a USB stack such as the Chumby, smart phone, <insert controller>. Of course, one could implement a USB stack on a microcontroller of one's choice (back to the significant software development effort).

                  It sounds like finding the smallest, cheapest, lowest power consuming, embedded computer with USB and on-board Wifi is the most straightforward approach.



                  On Sat, Jul 2, 2011 at 3:33 PM, Nathan Monson <nmonson@...> wrote:
                   

                  On Sat, Jul 2, 2011 at 3:16 PM, Weston Turner <wstnturner@...> wrote:
                  The data-sheet claims it "delivers up to 3 Mbps data rate for distances to 100M."

                  This is the very module I used.  Maybe it was something about the environments I tested in.  Both at the apartment and in the office, performance was just peachy until line of sight was lost.  There are plenty of other things chattering at 2.4GHz in both places, and metal cubicle frames can't be too good.
                   
                  Using Wifi will give you much more bandwidth and flexibility but at a cost of a greater integration effort.

                  I'm not so sure about this in the case of the XV-11.  One hacker found out that the XV-11 does have a UART, but it runs at a slow speed.  The 'high speed' sensor data can't exit this port.  The USB port is the only way to access the laser.

                  Since you must have a USB host stack to get the laser sensor, the integration is already beyond just gluing things together.

                  But a Chumby with WIFI is almost the perfect fit.  The Chumby recognizes the XV-11 as soon as you plug it in, and a couple of lines of shell scripting connects it over WIFI.  That's some nice integration!
                   
                  Also, I have doubts about the actual bandwidth needs of the application. Surely some pre-processing could be applied to the data stream to make it sparser.

                  This is very true, and a good idea no matter what protocol is used.  The XV-11 seems to use verbose ASCII to send data, and quite a bit of the data is of dubious or unknown value, such as error codes and laser intensity values.

                  Of course, the Chumby makes that pretty easy too.

                  Cell phones are another interesting choice.  I met somebody who hacked his Droid (software) to be a USB host, so any Chumby code would work equally well on that platform.

                  Nathan


                Your message has been successfully submitted and would be delivered to recipients shortly.