|Surveyor Robotics Journal|
Sat, 31 Jan 2009
Google Android G1 + Surveyor SRV-1
Just got our first live video running from the SRV-1 on an Android G1. Above is a screenshot of the emulator running the video feed, but this is running on the actual G1 phone as well. Motion control for the robot is next. More details to follow.
ps: huge thanks to Jeffrey Nelson, creator of the Forknife Robot Wed, 28 Jan 2009
adding a GPS to the Surveyor SVS / SRV-1
Last week, I added a Locosys LS20031 GPS to my SVS setup. At the moment, it is using UART1 at 57600 baud (standard configuration for the LS20031 from Sparkfun) on the left Blackfin, but plan to use GPIO pin 31 or 32 with a software-only UART when we have a chance to finish that code. The latest firmware (srv-blackfin-012509) has a command "$g" to dump the data from the GPS to the console, but one line in main.c has to be uncommented in order for this to work -
//init_uart1(); // use UART1 with GPS until GPIO soft-UART is ready
The setup is quite simple. I soldered a 4-pin header to the LS20031 as shown here -
and then ran a lead from the left-most pin on the GPS to pin 1 (3.3V) on the 32-pin header, a lead from the right-most pin on the GPS to pin 2 (GND) on the 32-pin header, and a lead from the pin next to GND on the GPS to pin 6 (UART1 RX) on the 32-pin header.
Finally, I put a block of non-conducting foam under the GPS to isolate it from the SVS.
This same setup works with the SRV-1. Only work remaining is to parse the incoming NMEA strings and move the RX line from UART1 to a GPIO pin once the soft-uart code is ready.
As mentioned on the robotics forum, to make this easier, we will be adding a 4-pin header for the GPS and another header for a Honeywell HMC6352 compass on the next version of the SRV-1 radio/motor board, so no wiring will be required.
Fri, 09 Jan 2009
new console for Surveyor SVS (stereo vision system) on Windows
Andrew Hutchinson of LochTurret.com has developed a very nice console for the Surveyor SVS using Delphi. The console supports the SRV-1 motor controls, lasers, image processing functions, range, stereo anaglyph rendering, etc, with provision for custom-designed functions as well.
The console can be downloaded from -
Thu, 08 Jan 2009
just moved surveyor.com to a new server
This should resolve the significant performance issues we had with the previous server, especially on the robot forum. The transition was reasonably easy, but it is likely that we broke a few links in the process. If you see any problems, please notify us by email at email@example.com. Thanks !
Sun, 04 Jan 2009
Robot NASCAR - modifying a stock r/c car for robot control
With the new year, I was reminded of the Robot Magazine RoboRacer project and my HPI Micro RS4 1:18 scale r/c car that had been gathering dust for 2+ years. I took a stock SRV-1 Blackfin robot controller, added connectors to the switched battery power and 2 servo signals, and mounted the Blackfin with little trouble to the RS4.
This is stock photo of an unmodified RS4 -
Here is my RS4 with the r/c receiver removed, ESC (speed controller) moved toward the front of the frame, and SRV-1 Blackfin board set with Wifi radio mounted on standoffs -
The underside of the SRV-1 radio/motor board with pins 5 and 6 of the expansion header connected to the servo input signals for steering and speed control -
The RS4 is controlled via Wifi connection to the same java console we use to control the Surveyor SRV-1 Blackfin Robot, though a joystick control would be much more effective than the button commands shown below. For those familiar with the SRV1Console, the only change was a redefinition of the 'M' motor commands in srv.config to 's' servo commands, e.g. changing
Next step is autonomous control. A race track is not complicated visually - the processor needs to differentiate between ground and walls, and eventually be able to detect other cars on the same track. Orientation and track position are more complicated - we might want to add road signs for visual marker beacons and perhaps use different color markings on the left and right walls. The design of the track will "steer" some of the decisions on the best approach to image processing. This is a cool project, and it would be nice to see it gain some "traction".