|Surveyor Robotics Journal|
Thu, 29 Nov 2007
Building SRV-1 Blackfin firmware on Windows
I originally assumed that the SRV-1 Blackfin firmware could only be compiled under Linux, but found out from one of our customers that there is a current Windows version of the Blackfin toolchain. After installing the toolchain and modifying the Makefile slightly, I was able to successfully build and test the srv1.ldr firmware image from Windows XP. I'll look forward to hearing from others who want to try this on Vista.
A discussion of the procedure for building the SRV-1 firmware is found here on the Surveyor Robotics Forum - http://www.surveyor.com/cgi-bin/yabb2/YaBB.pl?num=1196352263
Sun, 25 Nov 2007
UAV installation of SRV-1 Blackfin - Part 1
We are working with new quad rotor called the "X-3D-BL Scientific" from Ascending Technologies GmbH in Stockdorf, Germany, with the concept of integrating the SRV-1 Blackfin camera and radio board with the UAV flight controls.
Interface is relatively simple - the X-3D-BL has a very capable onboard inertial measurement unit integrated with the brushless motor controls, so the interface between Blackfin and UAV is a simple 38kbps UART. To power the SRV-1 boards, we are connecting directly to the UAV battery pack (11.1V LiPoly).
We have mounted the radio board with a single standoff in place of the original X-3D FM antenna, after drilling a hole in the radio board and clearing a trace that goes to the unused Zigbee header. We added a simple antenna bracket, so everything is securely mounted.
We have made some simple indoor flight tests to verify that there are no issues with balance or radio interference. Image transmission through the WLAN from the SRV-1 Blackfin boards is working fine, and the UAV flight is very stable.
Next step is to run the flight controls from the SRV1Console (with some redefined buttons) through the Blackfin to the X-3D via the UART. One short-term goal is to capture video through the SRV1Console video archive function. After than, an ultrasonic ranging module will be added to the system for ground clearance measurement and control. Beyond that, we have a very long list of projects, including servo control of the camera tilt angle. However, our near-term objective is to develop the necessary firmware to support customers who are building the SRV-1 Blackfin boards into UAV applications. Plus, this is really fun !
Mon, 19 Nov 2007
SRV-1 Blackfin robot setup instructions
A preliminary set of setup instructions for the SRV-1 Blackfin robot have been posted here - http://www.surveyor.com/blackfin/SRV_setup_bf.html. Check back from time to time as we update the instructions, and please contact us via email (firstname.lastname@example.org) to note errors or confusing instructions.
Sun, 18 Nov 2007
SRV-1 power consumption
We noticed that our prototype robots seemed to be running a long time between charges, so we finally took some measurements. We found that the robot was drawing only 220 milliamps from the 7.4V Li-ion battery pack while stationery but fully operational (i.e. transmitting video frames at full speed). With the robot moving on a flat surface, the overall consumption jumped to 400 milliamps.
As the robots are assembled with Li-ion battery packs rated at 2000 mAh (milliamp-hours) capacity, it is reasonable to expect operating time of 4-8+ hours off a single charge, depending on the amount of robot movement. This is comparable to the power consumption of the original SRV-1 with ARM7 and Zigbee.
In fact, the Blackfin and WLAN draw more current than the ARM7 + Zigbee combination, but the overall system efficiency is the same because we're using a Recom switching regulator to convert the battery's 7.4V to 3.3V with 90% + efficiency. The original robot used an inefficient LM7805 linear voltage regulator to dissipate the voltage drop from 7.4V to 5.0V as heat.
There are a variety of power save features available in both the Blackfin processor and the Matchport WLAN module that we have yet to explore. Just dropping the clock speed of the processor can make a substantial difference in consumption, and there are some similar adjustment that can be made to the Matchport. At the moment, the power consumption levels are not an issue, but we look forward to eventually experimenting with some power optimizations.
Updated support for SRV-1 Blackfin in RoboRealm
Roborealm.com has released an new version of software that includes updated support for the SRV-1 Blackfin robot and camera board. A description of the update is found here -
along with a tutorial on object tracking here -
For those not familiar, RoboRealm is a very popular Windows-based programming tool that combines an extensive suite of vision processing modules with programmable robot control. The software is well supported and it's free !
Tue, 13 Nov 2007
First robot shipments; connector changes; firmware changes ...
First production SRV-1 Blackfin robots shipped to customers last week, and we're starting to work our way through the order backlog. It will likely take at least a month to catch up with existing orders and reservations, so we are grateful for everyone's patience. Please note that there has been a price increase on the robots to $465 for orders / reservations received after November 8, but the original price will be honored for the early orders and reservations.
For users who are anxious to start work with the Blackfin processor sooner, we are staying reasonable current with orders for just the SRV-1 Blackfin Camera board and WiFi radio board. The Blackfin PWM counters can be configured to drive other servo-style motor controllers, and we will work with users who want to interface the Blackfin to other motorized bases. We are also exploring the possibility of cooperation with other robot base manufacturers, since that is the main bottleneck in our production process.
There is one minor change relative to production configuration of the processor cards. While the stacking connectors work well for the radio and motor control boards, they are very difficult to solder to the processor card because of the amount of copper ground plane. Also, it is unlikely that boards will be stacked on top of the processor card, as this blocks access to the JTAG connector, jumper headers, and the possibility of orienting the camera parallel to the processor. As a result, we're going to replace the stacking connector on the processor card with a single 32-pin male header that mounts downward from the bottom of the card. The only functional impact is that boards will no longer be able to be stacked on top of the processor, but as noted, that's an unlikely configuration anyhow.
With regard to SRV-1 Blackfin firmware, we are now configuring the processor card to directly boot the SRV-1 firmware instead of going through u-boot. This greatly simplifies the issue of u-boot's autoboot sequence getting aborted by incoming data, and the SRV-1 firmware can start immediately with a 921kbps interface. Also, we have added the ability to upload boot images in the SRV-1 firmware via XMODEM-1K file transfer protocol. Enabling this transition is a new set of software tools from the Analog Devices uClinux group that supports the LDR boot file format. For users who want to use u-boot and linux, it's easy enough to upload the u-boot loader and overwrite the SRV-1 firmware, plus srv1.ldr as well as u-boot.ldr are compatible with the Blackfin's UART boot mode. In time, we will create a useful roadmap that shows how to navigate these options, but be aware that the Blackfin provides a rich set of configuration options.
One last note - we have posted another linux test image for the SRV-1 on http://www.surveyor.com/blackfin/. This version has Python built in. Our goal is to create something similar to the Myro robot environment onboard the SRV-1 using these tools, and we already have at least one developer working in that direction.