|Surveyor Robotics Journal|
Sat, 31 Mar 2007
SRV-1 WiPort update - successful test of prototype PCB's
We received the prototype batch of circuit boards last week, and installed WiFi on several robots after initial testing.
One of the robots shipped to Lantronix the same day for display next week in their booth at the Embedded Systems Conference in San Jose - you can look for them in booth # 2410.
We kept one of the WLAN prototypes for testing, and everything worked as expected, though we found that we could actually run the module at 921kbps without hardware flow control as long as the JPEG frame sizes remained below 4kbytes (160x128 resolution). After discussing this further with Lantronix engineers and considering some schemes for software flow control, we've decided to modify the circuit board slightly to add a buffer overrun detection signal which will allow us to run the modules full speed at 921kbps, even with larger JPEG frames (up to 640x480 resolution).
We will test the modified circuit shortly, and then update the PCB layout and order a production run of boards. Realistically, we should have first completed modules ready to ship by April 13.
As regards pricing, we will offer the SRV-1 WiPort expansion board for only $15 plus freight to existing SRV-1 owners who supply their own WiPort module and antenna - this includes the voltage regulator, connector, mounting hardware, LED's and passive components - please note that this cost does not include the WiPort and antenna. This offer only applies to existing SRV-1 owners, as it is less than our actual production cost for the expansion board.
The cost of a complete SRV-1 WiPort radio, including expansion board, WiPort and antenna, will be $150 plus freight, and education discounts will apply.
The cost of a complete SRV-1 robot with battery charger and WiPort radio (i.e. SRV-1 robot starter kit with WLAN) in place of Zigbee radios will be $435 plus freight, and likewise, education discounts will apply.
If you are interested in any of these configurations, please contact us via email - email@example.com.
Thu, 29 Mar 2007
Report from 2007 AAAI Spring Symposium
Just back from a 3-day conference at Stanford University with university robotics educators discussing curriculum and tools for autonomous educational robotics - http://www.cs.hmc.edu/roboteducation/. As a large percentage of our robots are going to educators and researchers, I was hoping for some first-hand feedback on user requirements, how our robots fit, and what we need to change. The information gathered at the conference significantly exceeded my most optimistic expectations, it will take some time to fully digest what I learned, I found that we're generally on the right track, though we need to focus some effort on ease of use, especially with regard to software installation.
Concluding the symposium was a friendly competition between different participants who could choose different robots to work with to accomplish a specified task. The particular task was to program the robot to move away from a starting point for 30 seconds while avoiding obstacles, then turn around and try to try to get as close as possible to the starting point in 30 seconds. Robots used included the Parallax Scribbler with Myro, a variety of iRobot Creates with XBC Botball controllers, a CMU TerK, and the SRV-1.
The team from University of Kent, Canterbury UK, modified the SRV-1 firmware on the fly to run their "Transterpreter" - www/transterpreter.org, executing an onboard program in the occam-pi programming language, and building a three-layer Brooks-style subsumption architecture to accomplish the specified task. The Transterpreter is a particularly powerful architecture for programming concurrency, even on a simple processor such as the ARM7. Here's a picture of the flow diagram from which they based the code -
and a picture of the team crunching code with less than 20 minutes to competition deadline -
Here's the robot in action -
and a photo of the winners with their spoils -
Matt Jadud, Christian Jacobsen, and Jon Simpson from University of Kent
Our congratulations and thanks to the winners. We look forward to having them explain exactly what a three-layer Brook-style subsumption architecture is, and especially more about Transterpreter and occam-pi, as they plan to continue working with the SRV-1.
Thu, 22 Mar 2007
"Myro" - new Python-based support for SRV-1
A test version of "Myro" support for the SRV-1 is now available from IPRE (Institute for Personal Robotics in Education), a joint effort of Bryn Mawr College, Georgia Tech and Microsoft Research.
Myro is a new framework for programming robots. It is written in the language Python and designed for use in Introductory Computing courses. A full description of Myro is found here -
A full setup of Python under Windows, including Myro version 1.0.3 is found here -
And a description of Myro support for the SRV-1 is found here -
Python is a very nice programming language for robot control and the Myro functions are easy to use. Please give this a try and let us know if you find any problems. Mon, 12 Mar 2007
WiFi decision made - we're going with the Lantronix WiPort for the SRV-1
After evaluating a variety of alternatives and getting some good technical input from support engineers at Lantronix, we have decided to move ahead with the layout of a WiFi expansion card for the SRV-1 based on the Lantronix WiPort 802.11b/g WLAN module.
The card will measure 2.5" x 1.5" and will mount to standoffs that extend from the mounting holes on the current SRV-1 controller, in the same manner we used for the prototype installation of the WiPort as shown here - http://www.surveyor.com/cgi-bin/robot_journal.cgi/2006/09/02#047.
It will include 2 LEDS (power and WLAN activity), a high-efficiency Recom 3.3V switching regulator, a mounting hole for the WiFi antenna, and a high density connector for attaching the WiPort. Interface to the SRV-1 will be quite simple - +V lead to the battery power, and RX / TX signal leads that will plug into the existing Zigbee radio connector. The board will be grounded through the chassis.
By designing the card with a switching regulator, the power consumption of the WiPort will be reasonable - though the module uses 200-450mA at 3.3V, it will only be drawing 90-200mA from the battery. That is approximately double the current draw of the Zigbee radios.
We should have prototype WiFi expansion cards in 2-3 weeks, with production boards following shortly thereafter. We plan to continue support for the Zigbee radios, but will offer WiFi as an alternative. We haven't worked out pricing yet for the WiFi option on future SRV-1 orders, but there will be a special offer to existing SRV-1 customers. Those who are interested should contact us at firstname.lastname@example.org
Tue, 06 Mar 2007
WiFi ... the saga continues
If we had this to do over again, we should have committed to supporting the Lantronix WiPort after successfully testing it back in September. It still has the issues of requiring a special connector / breakout board and 3.3V regulator, so while the module costs $129, the extra components bring the total up to about $150. However, the WiPort works, the data rate is good, and it is in volume production. Until finding an alternative module that exactly matches our spec (5V, internal antenna, 4-wire 921kbps interface, sub-$80 cost, volume production), the WiPort would have satisfied most customer requirements for WiFi functionality. Of course, that is the benefit of having perfect 20:20 hindsight.
There are some new products coming to market in the next 3-6 months that will match our requirements, but we need to offer an interim solution. At this point, the most likely scenario is that we will build an adapter board with 3.3V voltage regulator to host the WiPort, or we will offer a lower data rate module such as the ConnectOne Socket iWiFi (230kbps) or the DataHunter Mini-B (115kbps) that don't require an adapter board. We have actual samples of the ConnectOne and DataHunter modules in route, so I am hopeful we can make a final decision on this shortly.
Sun, 04 Mar 2007
RoboRealm machine vision software adds support for SRV-1
The SRV-1 can now be directly controlled from RoboRealm, a very popular Windows-based machine vision software package for robots. The RoboRealm extensions for SRV-1 allow creation of scripts that combine image processing on live video feeds from the robot, e.g. color filtering, blob detection/tracking, edge detection/outlining and feature extraction, with decision processing and robot motion control, making it easy to create behaviors such as object location and tracking, obstacle avoidance, motion detection, notification, etc, with a web interface, and control can be scripted from C/C++, Python, Java, C#, Lisp, Visual Basic, WScript and COM through the RoboRealm API.
An overview of RoboRealm's support for SRV-1 is found here - http://www.roborealm.com/help/Surveyor_SRV1.php.
A tuturial using VBScript with RoboRealm and the SRV-1 for wall following is found here - http://www.roborealm.com/tutorial/Surveyor_SRV1_maze/slide010.php.
video from tutorial of SRV-1 roaming under control of RoboRealm
RoboRealm software is free (it's supported by website advertising) and very capable. There is an active forum and excellent technical support. Make certain to give this a try, and let us know if there are any issues either via the Surveyor Robotics Forum or the RoboRealm User Forum.