Prototyping and deployment made easier by design
Tom Shelley reports on software that speeds the development and prototyping of complex control strategies
Software originally developed to turn desktop computers into virtual instrumentation systems has now evolved into a way of prototyping complex control systems – testing them first with software models, then hardware. Finally they are deploying onto custom boards with 32-bit microprocessors for use in the real world.
“Building any kind of software and hardware application is getting more and more complex,” says John Pasquarette, software marketing director of National Instruments. “Historically, our software was used to test, but now it is typically being used to prototype designs with software that can then be deployed.”
LabView 8.20, the latest version of the software, includes an FPGA module that includes new machine monitoring functions for implementing filters, alarms and measurements so that designers can build FPGA-based machine protection systems. At the software’s launch, the company demonstrated an inverted pendulum, modelled in software. This is a notorious control problem in which the pendulum has to be swung into a vertically upright position and maintained there, by moving the carriage supporting it from side to side.
The demonstration included a haptic control knob interface in which the inertial forces produced by the swinging pendulum could be felt by the person turning the knob to move the carriage to and fro. Swinging the pendulum up and keeping it more or less vertical is extremely difficult, but modern control algorithms can achieve this with relative ease.
In a real world situation, devising a control algorithm, for say, an unstable and thus more highly manoeuvrable aircraft, the control algorithm would first be tested against the computer model, then against real world hardware, and then deployed to a dedicated control computer for use in the field. Real world problems are usually much more complex than a simple inverted pendulum, and it is in such cases that the graphical LabView interface and its various tools to automatically generate blocks of code come into play.
The company says that in the latest software release, the execution speeds of PID control loops have been increased by up to 14 times and execution speeds of the Simulation Module by up to nine times, making it possible to develop and execute control models with more than 1,000 nodes.
Control algorithms can either be generated using LabView, or imported from a MathWorks Simulink environment. In addition, with a new External Model Interface in the Simulation Interface Toolkit, it is possible to use values from third party models generated in Dynasim or Plexim software.
Designers can implement FPGA based controllers on plug-in boards on a standard desktop PC or in modular NI CompactRIO or PXI hardware platforms. For the first time, it is also possible to target control algorithms directly to custom board designs using 32-bit microprocessors using the LabView Embedded Development Module.
“Where we have most use right now is in academia but we are seeing a lot of interest from industry, from makers of white goods – especially washing machines – and manufacturers of medical devices,” said Pasquarette.
Some indication of ease of use may be gleaned from the fact that the same technology underlies the LabView Toolkit for the Lego ‘Mindstorms NXT’ robotics platform which is targeted at children aged 10-14.
National Instruments
Pointers
* LabView has expanded out of virtual instrumentation into industrial control
* It speeds the design, simulation and testing of complex control strategies that can then be deployed onto dedicated 32-bit microprocessor-based hardware