AHS International 59th Annual Forum 2003

CORBA in Control Systems

Presented at the OMG Workshop On Distributed Object Computing For Real-Time and Embedded

Systems, Washington, DC, 14-17 July 2003

Dr. Shahzad Aslam-Mir, Prism TechnologiesDr. James L. Paunicka, Boeing

Eric J. Martens, Boeing

AHS International 59th Annual Forum 2003

• Sam’s slides:– Intro– Control Systems– Importance of CORBA in Control– Current schools of thought on CORBA applicability in control systems– Historical perceived limitations on CORBA use in control loops– Case studies

AHS International 59th Annual Forum 2003

Flight Control Example• Standards-based CORBA middleware used in live flight of a

commercial rotorcraft UAV in May 2002– Flight test was collaboration between Boeing and Georgia Tech School of

Aerospace Engineering» Boeing – embedded software architecture, integration of software on

embedded processing platform» Georgia Tech – flight vehicle, vehicle control software

AHS International 59th Annual Forum 2003

Flight Control Example• Flight test was part of the DARPA IXO (Information eXploitation Office) SEC

(Software Enabled Control) program

– Significant technical direction from Air Force Research Laboratory» Information Systems, Advanced Architecture

and Integration organization» Air Vehicles, Control Systems Development

and Applications organization– Leverages OCP (Open Control Platform) being developed on the SEC program

» Middleware platform that adapts Boeing Bold Stroke software technology to the domain of flight vehicle control

» ACE/TAO» Run-time optimizations to support flight vehicle control» API for flight vehicle control applications (“Controls API”)» Developed by Boeing-led team that also includes Georgia Tech, University of

California-Berkeley, and Honeywell

AHS International 59th Annual Forum 2003

Background

• SEC funds two technology areas

– Control technology for flight vehicles

» Fixed-wing and rotorwingUAVs

» Multiple research teams from industry and academia

– Enabling Software technology

» Adapt Boeing Bold Stroke software technology to the domain of flight vehicle control Hardware (CPU, Memory, IO)

Operating System & Board Support PackageACE & RT CORBA (TAO)

OCP Middleware ExtensionsControls API

OnlineModel Controller Gnd

Station

Hardware (CPU, Memory, IO)Operating System & Board Support Package

ACE & RT CORBA (TAO)OCP Middleware Extensions

Controls API

OnlineModel Controller Gnd

Station

Open Systems HardwareOpen Systems Software

OCP Middleware Download

Software Application Components

AHS International 59th Annual Forum 2003

Commercial UAV Flight Vehicle

• Yamaha R-Max autonomous helicopter– Couple hundred pounds, – 10-foot main rotor diameter

• Fitted with open systems avionics platform for SEC program experimentation

– Sensors» IMU (Inertial Measurement Unit), GPS, sonar for altimeter, magnetometer

for compass– Actuators

» Throttle, main rotor, tail rotor– Comms

» Wireless ethernet» Wireless serial link

– Onboard compute platform » Single 266-MHz Pentium II processor

AHS International 59th Annual Forum 2003

R-Max Vehicle

AHS International 59th Annual Forum 2003

Autonomous Helicopter Electronics

R-Max Actuators(blades & throttle)

Yamaha ControlSystem (YCS)

Yamaha R/Max Sensors

Open SystemsAvionics

YCS Actuator Cmd’s

GIT Actuator Cmd’s

Actuator Command PathSwitch (Pilot & YCS controlled)

Open Systems Equipment Bay

Wireless Ethernet Link

Wireless Serial Link

Yamaha HardwareSide

R/C Link

IMU

MagnetometerSonarGPS

YCS System & Sensor Data

R/C PilotControl Box

AHS International 59th Annual Forum 2003

Flight Test Embedded Software Architecture

• Major Elements– Lowest level – VxWorks RTOS and appropriate BSP (Board Support

Package)– Middleware level – Open Control Platform from SEC program– Application software level – multiple components written by Georgia

Tech• Run-time configuration of CORBA-based software

– Middleware triggered execution of multiple software components with EC (Event Channel)

» 100 Hz operation• Triggered the start of inner-loop control processing with the arrival of IMU data

– Middleware mediated I/O among the various aircraft sensors, flight control actuators, and multi-level control loops

– Implemented a software reconfiguration in flight» Neural net adaptive controller switching to a conventional inverting

controller

AHS International 59th Annual Forum 2003

Flight Test Embedded Software Architecture in Context

GPS (Signal A at 5 Hz & Signal B at 5 Hz ... not at same time)

IMU (100 Hz)

Magnetometer (20 Hz)

sonar (40 Hz)

receiver commands (40 Hz)

Vehicle Health (? Hz)

RMAX Attitude sensors (200 Hz)

Sensors SerialInterface

Vehicle SerialInterface

Controls API Input Port

Controls API Output Port

RMAX Actuator demultiplexer

Actuator SerialInterface

NavControlOutputPort NavControlInputPort

NavDataOutputPort

NavDataInputPort

IMUTrigger

50 Hz

100 Hz

I/OComponent

DataLink Interface Ethernet “Serial” PortSerial port

Ethernet “Serial” Port

Serial port1 Hz & 10 Hz1 Hz & 10 Hz

Input datalink portsread @ 100 Hz

m0 written at 10 Hzm1 written at 1 Hz

Navigation ModuleComponent

ControllerComponent

50 Hz

<NavRunCommand>

Global DataStorage

<ControlRunCommand>

<Signal Name> - passed signal

AHS International 59th Annual Forum 2003

Flight Test Embedded Software

• Multiple Application Components, including

– I/O Handler– Navigation Processing

» Also handles data link updates– Controller Processing

• OCP Frame Manager launches 100-Hz loop after 16-byte IMU burst

– Pushes an event to start I/O Handler

• Other components also initiated in 100-Hz frame with event pushes

• Middleware-Based Reconfiguration accomplished with OCP activating and deactivating different controllers

WindView Plot of OCP-Based Application

AHS International 59th Annual Forum 2003

Flight Test Timeline

• R/C (Radio Control) pilot performed take-off (and subsequent landing) using baseline Yamaha flight control system

• While in flight, rotorcraft switched to open systems research flight control system

– With RT-CORBA-based flight control program

• Transition between vehicle controllers (neural net to conventional inverting controller) triggered by ground station command and accomplished by middleware

Video from Flight Test

AHS International 59th Annual Forum 2003

• Sam’s slides:– Analysis of where we are– Conclusions