Posted by SLUGS Team on January 22, 2009 at 4:13pm
The Santa Cruz Low-cost UAV GNC System (SLUGS) is a high-end UAV Autopilot that is currently under development by the Autonomous Systems Lab at the University of California Santa Cruz. Although not ready for prime time, the project's website shows some significant progress towards a fully functional platform.The SLUGS autopilot is heavily geared towards research in small to medium sized UAVs. Great care has been taken to assure that it has enough processing power for moderately complicated control tasks and at the same time is easily reprogramable via Simulink. This effectively allows to rapidly iterate from simulation to implementation with no intermediate coding.The developed architecture physically decouples sensor integration and INS filtering from the control algorithm and communications by using two 33Fj256MC710 dsPIC Digital Signal Controllers (DSC) interconnected via a high-speed Serial Peripheral Interface (SPI) bus. The SLUGS autopilot has been designed to be modular and extendable in order to extend the sensor and peripheral suite as the need arises via CAN bus daughterboards.The Ground Station Software has been developed to decode the incoming telemetry messages from the SLUGS AP. It also works as a Hardware-In-the-Loop interface to Simulink and is the central point to configure the Autopilot.One of the main features of the SLUGS AP is its ability to interact with virtually any Simulink model via a Hardware-in-the-Loop Simulator. There is a predefined protocol to send simulated sensor readings to the AP, as well as a protocol to receive control commands back. But aside from that, one can make the HIL simulator as simple or as complex as one needs. Currently we have a six-degree-of-freedom model of our UAV, and an engine model. Some work is underway to include an atmospheric and weather model.We are currently working very hard to write some sort of documentation and getting all the code ready for public release under the MIT Open Source license and under GIT version control system. We expect this to be ready around April this year.
@Noth666, we will NOT sell anything, not now and not in the foreseeable future, so not sure what you mean by "available in Europe." The source code (and binaries), Schematics, Layouts and Gerbers will be available so you can roll your own.
@Varga András: Version 1, yes, everything has been hand soldered by one of the team members. But, version 2 will be much harder to solder by hand. We are planning on using solder paste and a skillet. Checkout Sparkfun's tutorial on doing that, we have made some preliminary tests and have had great results.
Is it possible to mount the whole stuff on the PCB manually (for example, I mean there's no BGA packaging)? (For someone who is pretty good at soldering)
Yes, programming from Simulink is a very nice feature, and you are correct, ti speeds up the whole development and implementation process.
The first version was about $600 which includes 4-layer board, GPS and components. We have mixed emotions about next revision. We are thinking about going to the more expensive Analog Devices "cube" which is $600 by itself, but apparently much better. But of course that makes the entry level so much harder. We have not decided on that. But, we are considering cheaper and easier magnetometers and if we don't go with the "cube" then we expect the price to come down to about $450, since we will be eliminating many "extra" components we had on the first version. The board designer and layout expert is working on that as I write this. We will certainly blog about it.
Very impressive, I like how it's programmable straight from Simulink, I imagine that can speed up control algorithm development quite a bit. Do you have a rough estimate on the price of the hardware?
Comments
-Brad
@Varga András: Version 1, yes, everything has been hand soldered by one of the team members. But, version 2 will be much harder to solder by hand. We are planning on using solder paste and a skillet. Checkout Sparkfun's tutorial on doing that, we have made some preliminary tests and have had great results.
The first version was about $600 which includes 4-layer board, GPS and components. We have mixed emotions about next revision. We are thinking about going to the more expensive Analog Devices "cube" which is $600 by itself, but apparently much better. But of course that makes the entry level so much harder. We have not decided on that. But, we are considering cheaper and easier magnetometers and if we don't go with the "cube" then we expect the price to come down to about $450, since we will be eliminating many "extra" components we had on the first version. The board designer and layout expert is working on that as I write this. We will certainly blog about it.
-
1
-
2
-
3
of 3 Next