(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
`
`AIAA Guidance, Navigation, and
`Control Conference and Exhibit
`14-17 August 2000
`Denver, CO
`
`AOO-37046
`
`AIAA-2000-4057
`
`HIERARCHICAL CONTROL SYSTEM SYNTHESIS
`FOR ROTORCRAFT-BASED UNMANNED AERIAL VEHICLES
`
`David Hyunchul Shim*, Hyoun Jin Kirn*, Shankar Sastry**
`
`ABSTRACT
`
`This paper introduces the development of multiple
`number of Unmanned Arial Vehicle (UAV) system
`as a part of BErkeley AeRobot (BEAR) project,
`highlighting the recent achievements in the design
`
`
`and implementation of rotorcraft-based UAV
`(RUAV) control system. Based on the experimental
`flight data, linear system model valid near hover
`condition
`is
`found by applying
`time-domain
`numerical methods to experimental flight data. The
`acquired linear model is used to design feedback
`controller consisting of inner-loop attitude feedback
`control, mid-loop velocity feedback control and the
`outer-loop position control. The proposed vehicle-
`
`
`
`
`level controller is implemented and tested in
`Berkeley UAV, Ursa Magna 2, and shows superior
`hovering performance. The vehicle level controller is
`integrated with higher-level control using a script
`language framework to command UAV.
`
`1. Introduction
`
`The rapid development of the sensor, communication
`and control technology in the last few decades has
`made autonomous vehicles
`powerful. And the need to alleviate the actual human
`participation in order to save human efforts and avoid
`hazards has been ever increasing in numerous fields.
`As a result, autonomous vehicles are about to become
`part of reality in many applications.
`
`smaller and more
`
`BErkeley AeRobot project aims to organize multiple
`number of autonomous agents into integrated and
`intelligent systems with reduced cognition and
`
`control complexity, fault-tolerance, adaptivity to
`
`'Graduate Students, Department of Mechanical
`Engineering, University of California at Berkeley
`**Professor, Department of Electrical Engineering
`and Computer Science, University of California at
`Berkeley
`{hcshim,jin,sastry}@robotics.eecs.berkeley.edu
`Copyright © 2000 by David Hyunchul Shim.
`Published by the American Institute of
`Aeronautics and Astronautics Inc. with
`permission.
`
`changes in task and environment, modularity and
`scalability to perform complex missions efficiently.
`Figure 1 illustrates one scenario in which a fleet of
`low altitude Unmanned Arial Vehicle fly over a
`suspected area
`looking for hiding ground-based
`enemies, possibly Unmanned Ground Vehicles
`(UGVs). UAV at higher altitude coordinates UAVs at
`lower altitude and reports the current status to the
`remote base via wireless communication media. The
`remote base receives the mission states and issues the
`following-up commands.
`
`To realize this scenario, we need a number of
`cooperating UAVs that are able to navigate the area
`following waypoints commanded by the higher-level
`guidance system. They should be equipped with
`various sensors, which provide estimates of state
`variables and detect objects of
`interest, and
`
`controllers which generate the actuator signals
`
`
`
`
`required to follow the reference attitude and
`trajectory properly. And the guidance system needs
`to be able to process reference trajectory suitable for
`the mission. As for the ground-based agents, similar
`automated vehicles with navigation capability are
`required and these are available commercially.
`
`Figure 1. Mission coordination of multiple agents
`
`The rotorcraft suits our scenarios particularly well
`due to its versatile maneuver-ability such as vertical
`take-off/landing, hovering, sideslip, pirouette and so
`on. The rotorcraft based UAV can be operated in
`relatively smaller space by vertical
`take-off and
`landing. It is also able to hover over or track a target
`agent below at matching low speed.
`
`1
`American Institute of Aeronautics and Astronautics
`
`Parrot 1021, Page 1
`
`

`
`(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
`
`Compared with the mission-level guidance tasks, the
`actual UAV navigation and control problem is not
`fully investigated and the diverse research activities
`are ongoing[2,3,4,5,6]. One of the daunting tasks in
`constructing a UAV system is to obtain high fidelity
`
`model to design
`flight controllers upon. The
`the model
`traditional approach of borrowing
`developed for full-size helicopters often fails to
`account for the unique dynamics of RUAVs, which
`are commonly equipped with the servorotor system.
`Recently, Mettler[2] suggested a very effective model
`to account for the dynamics of servorotor. In this
`research, this model is adopted for the identification
`of the hovering and instead of frequency-domain
`identification, time-domain method is applied. The
`obtained model is used for multi-loop classical
`hovering control design and full-model linear robust
`control synthesis.
`
`Vehicle Control Language(VCL) is also proposed in
`this paper as an application and development tool for
`language system,
`RUAV systems. This script
`operating in a hierarchical client-server environment,
`integrates the vehicle-level control and the higher-
`level motion command.
`
`engine and can carry up to 20 kg. As the intermediate
`size, Bergen Industrial Twin helicopters are chosen.
`They are powered by customized two cylinder four
`stroke gasoline engine and offers 10 kg payload,
`which is ideal for carrying the basic navigation/
`control package and some other extra equipment such
`as vision processing board, camera system and so on.
`
`Figure 2. UAV fleet at UC Berkeley
`Kyosho Concept 60 Graphite has been used as the
`valuable testbed for prototyping the flight system
`design[6]. Classical SISO position/velocity/attitude
`controller has been designed and tested successfully.
`In this paper, similar approach is applied to Yamaha
`R-50 helicopter.
`
`The organization of the paper is as follows.
`
`
`configuration of
`Hierarchical architecture and
`A UAV is a vehicle integrated with mechanical and
`Berkeley UAV test bed system are explained in Sec.
`electronic components such as airframe, navigation
`II. In Sec. Ill, we present the dynamic model
`sensors, computers, batteries and other sensors,
`identified from Berkeley RUAVs, highlighting the
`performing autonomous tasks desirably with minimal
`effect of servorotor. In Sec. IV, we discuss how we
`intervention by remote human operator. The onboard
`design stabilizing control
`law and present
`the
`components can be categorized into the following: 1)
`experiment results. In Sec V, the novel approach of
`flight control computer (FCC), 2) navigational
`VCL is introduced and examples are given.
`sensors, 3) communication module, and 4) onboard
`power pack.
`
`2. Berkeley UAV System Architecture
`
`Berkeley UAV research test bed is established in the
`context of multiple number of UAV and/or UGV
`agents operations. For base airframe, BEAR has
`chosen four different sizes of model helicopters. Two
`Kyosho Concept helicopters were built and equipped
`with navigation and flight control systems. These 60
`class helicopters are powered by 0.90 cubic-inch two-
`cycle glow-plug engine, which can carry up to 5 kg
`as payload. At the other end of the UAV size
`
`spectrum, two Yamaha agricultural helicopters,
`Yamaha RMAXs, are adopted for their sufficient
`payload
`
`(>30kg) and
`reliable,
`consistent
`performance. Yamaha R-50,
`the predecessor of
`RMAX, is slightly smaller version and is powered by
`single cylinder water-cooled
`
`two-stroke gasoline
`
`Vision
`Computer
`vittti
`CoJor thicker
`
`Linux
`P266
`
`Nav Computer
`
`QNX
`P233
`85MB Flash Disk
`40MB RAM
`
`Figure 3. Onboard flight control system structure
`
`American Institute of Aeronautics and Astronautics
`
`Parrot 1021, Page 2
`
`

`
`(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)1 Sponsoring Organization.
`
`Name
`Kyosho
`Concept 60
`Bergen
`Industrial Twin
`
`Length
`
`1.4m
`
`Table 1. Berkeley UAV physical specification
`Weight
`Height
`Engine
`9.3kg
`(4.5kg+4.8kg
`avionics)
`
`0.47m
`
`OS FX91 2.8bhp
`
`1.8m
`
`0.6m
`
`7kg dry weight
`
`Twin Genoa
`Gasoline engine
`
`Yamaha R-50
`
`3.5m
`
`1.08m
`
`54kg
`(44kg+10kg
`avionics)
`
`Water cooled
`2stroke 1 cylinder
`gasoline engine
`
`Yamaha RMAX
`
`3.63m
`
`1. 2m
`
`60kg dry weight
`
`Water cooled
`2 stroke 2 cylinder
`gasoline engine
`
`Autonomy
`Boeing DQI
`NovAtel RT-2
`MediaGX233
`
`N/A
`
`Boeing DQI
`NovAtel RT-2
`Pentium233
`4 ultrasonic altimeter
`Digital compass
`Vision processor
`
`N/A
`
`Flight control computer is constructed using PC 104-
`compatible boards due
`to
`their
`industry-grade
`reliability, compactness and expandability. The main
`board is powered by Pentium 233MHz MMX CPU
`with 64MB RAM and 72MB FlashRAM. Serial port
`expansion board, counter/timer board, custom take-
`over board (TOB), and DC-DC conversion power
`supply board are "stacked up" on the CPU board via
`PC 104 for communication, servo control and power
`supply, respectively. As the heart of the navigation
`sensor, Boeing DQI-NP inertial navigation system
`(INS) is adopted. DQI-NP consists of a pack of solid
`state inertial sensors and digital signal processors
`with serial port. DQI-NP outputs the navigation
`solutions in proprietary message format via serial
`port. It needs periodic position update from external
`sensor to correct the position estimation error.
`
`Figure 4. One of Berkeley UAVs: Ursa Magna 2
`
`The global positioning system (GPS) used in this
`research is NovAtel RT-2, which has remarkable
`
`accuracy of 2cm. The flight control computer
`acquires the position, linear/angular velocity and
`attitude from DQI-NP, and high accuracy position
`
`estimate from NovAtel RT-2 via RS-232. It also
`relays
`the converted position estimate message
`packet from GPS to DQI-NP every second. Based on
`the acquired navigation data, FCC computes the
`control output for four channels: main rotor collective
`pitch,
`tail
`rotor collective pitch, main
`rotor
`longitudinal cyclic pitch, and lateral cyclic pitch.
`These control surfaces are actuated by commercially
`available servomotors, which accept PWM signal
`(14-21 ms period, 0.8-2.4 ms on duty) as the
`reference command. The output angle of servo rotor
`is proportional to the duty-on duration. The PWM
`signal is generated by Intel 8254 counter/timer chip.
`To ensure safety, a special circuit is added on the
`TOB to switch from FCC control to human pilot
`command by a toggle switch on the radio transmitter.
`Other channels of counter/timer board read the radio
`receiver output to log the human pilot's command,
`which has been proven to be extremely valuable for
`system identification and feedback-assisted flight.
`
`The communication module contains two 900MHz
`wireless modem cards and one 2.4GHz wireless
`Ethernet card. The type of communication device is
`chosen based on the mission type. The wireless
`modems are preferred for long range mission because
`of their superior range up to 20 miles. The drawback
`is the relatively slow throughput (<11.5kbps). One
`wireless modem is used for data communication and
`the other is used for reading the differential GPS
`broadcast data. In normal situation, 2.4GHz wireless
`LAN is preferred because of their high bandwidth (up
`to UMbps), versatility, and
`low power output
`minimizing the potential interference with sensitive
`GPS operation. Currently, wireless Ethernet system is
`
`used as the backbone of multi-agent system
`consisting of multiple number of UAV, UGV, the
`ship motion simulator and the ground station.
`
`American Institute of Aeronautics and Astronautics
`
`Parrot 1021, Page 3
`
`

`
`(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
`
`Ground station consists of a GPS base station and a
`portable computer connected with a communication
`device such as wireless modem or wireless Ethernet.
`Ground station monitors and stores the flight data of
`the UAV and also sends the navigation commands
`comprised of vehicle control language as explained in
`Sec. V.
`
`design of high-performance flight control system. In
`general, however, it is often a challenging task to
`perform system identification of a rotorcraft based
`UAV system due to its multi-input multi-output
`(MIMO), nonlinear characteristics, severe noise and
`disturbance, and wide flight envelop.
`
`As our first step, we attempted to obtain a linear,
`time-invariant model valid in near-hover conditions.
`A 6-degrees-of-freedom linear rigid body helicopter
`model augmented with first-order approximation of
`
`servorotor dynamics is given by a differential
`equation
`
`x = Ax + Bu
`
`u = [ua u, UH ur ]r
`
`b,, BM
`
`r,,, J
`
`L "„
`
`(1)
`
`(2)
`(3)
`
`Larger UAVs are equipped with an onboard vision
`processing unit (VPU) and a camera actuated by pan-
`tilt-zoom platform. VPU can track a target object
`with certain color and computes its coordinate based
`on the navigation data received from the onboard
`FCC via serial link. It can be accessed by its
`where
`independent wireless Ethernet for monitoring and
`debugging purposes. VPU serves the vital role for
`x = [M v p q <I> 0 als bls w r
`vision-based landing, ground object detection and
`map building.
`
`3. System Identification
`
`The acquisition of high fidelity system model of
`target UAV is a crucial step towards the successful
`
`body-coordinate velocity
`u, v, w:
`<&, 0, *F: roll, pitch, yaw angle, respectively
`
`p, q, r: roll, pitch, yaw rate, respectively
`flapping angle
`als,bls\
`
`0 g 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
`
`0 0 0 0 0 1
`
`-1
`
`0 0 0 0
`
`0 0 0 0 1 0 0-
`
`1
`0
`-0.0178
`0
`
`0 0 0 0 0 0 0 0
`
`"-0.0954
`0
`-0.2047
`-0.0836
`
`0
`-0.2221
`0.1521
`-0.0514
`
`0 0 0 0 0 0 0 0 0 0 0 0 0
`
`0 0 0 0 0 0 0 0 0 0 0 0 0
`
`A =
`
`B =
`
`-0.5912
`-2.4055
`
`1.9729
`-0.0993
`
`15
`
`2454
`0
`
`116.9520
`-46.9690
`0
`
`0 0 0
`
`0 0 0
`
`(4)
`
`0 0 0 0 0 0 0 0
`
`0 0 0 0 0 0 0 0
`
`0
`5.7974
`-4.4017 -46.959
`2.3394
`-5.4830
`
`0 0
`
`0
`
`0 0 0
`
`0
`-2.6645
`-2.6645 0
`
`0.5543
`-28.850 -121.20 -0.5377
`0.0746
`0
`0
`0
`
`-8
`0
`22.140
`67.74
`
`0
`£
`
`32.995 0
`142.50
`
`0
`0
`
`-g
`0
`0
`0
`00
`00
`0
`0
`0
`00
`00
`'
`
`Table 2. Eigenvalues of the identified helic
`
`Mode
`Phugoid 1
`Phugoid 2
`Roll
`Pitch
`Yaw
`Heave
`
`Value
`-0.4419±0.
`0.2796±0.(
`-1.3608±11
`-1.3001±8.
`-4.9507±10
`-0.52K
`
`American Institute of Aeronautics and Astronautics
`
`Parrot 1021, Page 4
`
`

`
`(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)1 Sponsoring Organization.
`
`feedback gyro system state .
`rft :
`One candidate model of eleventh order takes the form
`of Eq. (4) as suggested by Mettler et al [2]. One
`distinction of this model is the explicit account for
`
`the servorotor, which modifies the helicopter
`dynamics significantly. The most important role of
`servorotor is to slow down the roll and pitch response
`so that human pilot on the ground can control the
`helicopter with a remote controller. Heave dynamics
`is approximated by first order quasi-static model.
`This model yields linear, low-order approximation of
`the nonlinear high-order heave dynamics. For higher
`bandwidth controllers, the third or fourth order of
`model containing inflow and flapping dynamics
`should be used [7]. Yaw dynamics is inherently
`stable and modeled as first order system with
`reasonable fidelity. One special feature of the yaw
`dynamics is the built-in feedback action of yaw rate
`in the loop, which is provided by the built-in rate
`gyro amplifier/mixer. Even though uncompensated
`yaw dynamics is stable, the variation of the anti-
`torque of the main rotor continually perturbs the
`heading of the helicopter. The yaw rate feedback
`counteracts the torque by compensating the tail rotor
`collective pitch and left in the UAV system in case of
`manual flights. The gyro system model suggested by
`
`Mettler[2] is effective to account for the yaw
`dynamics. One deficiency of the model (4) is the
`absence of the cross coupling from yaw to sideslip
`and roll. It can be additionally parametrized in the
`model, but it turned out during the numerical process
`that the additional parameters are cumbersome to find
`because they appear as a product of parameters and
`the numerical process becomes singular.
`
`the
`
`Since the model is treated as a linear model, the
`nonlinearity of helicopter aerodynamics should not
`be excited by the excessive amount of control action.
`Hence, the appropriate design of control input signals
`for
`flight
`test
`is extremely
`important for
`identification of target flight dynamics. A number of
`experimental flights have been made to collect the
`flight data and the pilot input at 50Hz sampling rate.
`The control input consists of the combination of
`frequency-sweeping
`and
`random
`signals
`in
`longitudinal, lateral, yaw and heave channels in turn.
`At the first stage, the control in longitudinal and
`lateral channels are given simultaneously to capture
`the coupling between these axes while other channels
`are controlled to maintain constant altitude and
`heading. In the next stage, main rotor collective pitch
`or tail collective pitch is perturbed. Finally, control
`
`signals are issued in all channels to capture the cross-
`coupling term.
`
`Before processing the data, the angular rate signals
`are filtered by zero-phase noncausal discrete-time
`filters to filter out high frequency noise without
`introducing phase delay. The experiment result is
`processed
`using
`time-domain
`output-error
`minimization
`tool
`from MatLAB™ System
`Identification Toolbox™. The prediction error
`method (PEM) is an estimation algorithm, which
`seeks to a set of parameters minimizing the quadratic
`error between the predicted output and experiment
`data[8,9]. It should be noted that this method is
`
`extremely sensitive the initial guess of the to
`
`parameters and easily trapped in local minima of the
`parameter hypersurface. To obtain meaningful results
`other than some parameter set that blindly matches
`the time history, the following technique is devised.
`First, the angular dynamics augmented with rotor
`dynamics is identified using initial guess. Since the
`angular rate/rotor dynamics is known to be stable and
`the
`involved,
`small number of parameters are
`numerical solution converges to consistent solutions.
`Then the horizontal dynamics, i.e., the longitudinal
`and lateral dynamics with linear velocity terms u and
`v, are identified while fixing the angular dynamics
`parameters. This stage is rather challenging due to the
`unstable linear velocity dynamics. Shorter length of
`experiment data should be used
`to avoid
`the
`instability of the predictor and the divergence of the
`prediction error with the small mismatch of the initial
`condition and parameters. The solution is found after
`a large number of iterations using the experimental
`data from different time intervals. Separate from the
`longitudinal and lateral dynamics, the heave and yaw
`dynamics
`is
`identified
`in similar manner. The
`inherent stability of yaw and heave allows nice
`convergence of the parameters. Once these two
`subsystems are identified, they are combined as the
`full-model dynamics and then the cross-coupling
`terms are estimated. Finally, a small number of
`iteration is performed to recalibrate the parameters in
`the subsystems.
`
`4. Controller Design and Experiments
`
`Based on the identified model in Sec. 3, stabilizing
`control law is designed. As our first approach, we
`employ single-input, single-output (SISO) control
`structure. Classic control design approach has a
`number of advantages such as simple structure,
`straightforward design process and low computing
`
`American Institute of Aeronautics and Astronautics
`
`Parrot 1021, Page 5
`
`

`
`(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)1 Sponsoring Organization.
`
`load imposed on the FCC. On the contrary, it is
`limited by a number of disadvantages: it does not
`provide a systematic way to account for uncertainty,
`disturbance and/or noise. Moreover, it has limited
`capability to alleviate the coupling among channels.
`
`The proposed control system consists of three loops:
`innermost attitude controller, mid-loop linear velocity
`controller, and the outer loop position controller, as
`shown in Figure 5.
`
`The attitude dynamics of RUAV with servorotor
`
`mechanism has a unique dynamics, which is
`significantly different from the full-size helicopter
`without servorotor system[l]. The attitude dynamics
`of RUAV can be considered marginally stable if
`translational velocities are fixed at zero. On the
`contrary, the full-size helicopter without servorotor
`exhibits unstable attitude dynamics, which must be
`stabilized by angular
`
`feedback. The attitude
`controller proposed in this research only feed back
`the deviation of the roll and pitch angles from the
`trim condition and does not feed back the noisy
`angular rates p and q measured by solid state rate
`gyros. This approach yields a controller that is
`simpler and more robust to mechanical vibration. The
`adequate angular feedback gains for roll and pitch
`channels are determined to have acceptable response
`speed and damping by using root locus and step
`response.
`
`Figure 5. The proposed controller architecture using
`SISO multi-loop controllers
`
`The translational velocity dynamics of RUAV is
`almost identical to that of a full-size helicopter. This
`part of dynamics is unstable even with attitude
`feedback and requires velocity feedback. It can be
`stabilized with velocity feedback with constant gain,
`which is again found by conventional root locus and
`step response method.
`
`For position controller design, the system model (2)
`is augmented with three parallel integrators, which
`integrate translational velocities of x, y, z direction,
`
`respectively. The actual position
`integration,
`however, involves the coordinate transformation
`using Euler angles, which cannot be described within
`linear equations. Hence, the position control involves
`internal coordinate transformation to compensate for
`the heading change. The position gains are found by
`applying the similar methods described above to the
`augmented RUAV dynamics with velocity and
`attitude feedback.
`
`Apart from x-y directional dynamics, the vertical and
`the heading control require their own controllers.
`Although vertical and heading dynamics shows
`considerable coupling, conventional SISO control
`approach is maintained. These sub-dynamics show
`inherently stable response due
`to
`the natural
`equilibrium of the change of inflow and the generated
`lift. The vertical response can be further improved by
`introducing artificial damping by negative velocity
`feedback. For yaw rate response, the dynamics is
`already compensated enough by built-in gyro system.
`Still, the yaw angle response is marginally stable and
`the yaw tracking system can be built by simple
`heading error feedback with constant gain in its
`simplest form. The proposed controller based on fully
`decoupled SISO model is then tested on the full
`model, investigating the level of coupling when the
`loop is closed. The simulation showed that the closed
`loop system shows satisfactory performance.
`
`This simple structure of classical approach enables
`simple, but very effective control algorithm. In cruise
`mode, the velocity and attitude loops are closed for
`stabilization and tracking. When hovering over a
`certain spot is required, the outmost loop for position
`feedback is closed along with the inner loops. This
`algorithm is implemented in FCC and showed this
`idea actually works for real situation.
`
`The proposed control algorithm is implemented on
`the FCC. FCC is running on QNX real-time operating
`system. The onboard navigation and control software
`has two main processes and two auxiliary processes
`running concurrently, interacting with INS, GPS,
`servos, ultrasonic sensors, and vision computer.
`Process DQIGPS reads the GPS information and
`updates the INS using the GPS position estimate at
`IHz. Process DQICONT the INS
`
`
`measurements at
`lOOHz. These INS and GPS
`measurements are stored in the shared memory space
`
`
`
`reads in
`
`American Institute of Aeronautics and Astronautics
`
`Parrot 1021, Page 6
`
`

`
`(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)1 Sponsoring Organization.
`
`accessible to client processes. The control output is
`calculated using INS and GPS measurement at 46Hz
`and then sent to the registers of counter/timer chip,
`which generates a set of five PWM signals. The
`output to five servos can be switched from the signal
`from radio receiver or the computer generated PWM
`signal transmitted over optoisolators to guarantee
`stable operation of built-in Yamaha controllers and
`onboard computer systems.
`
`45 45.2 45.4 45.6 45.8 46
`
`0
`
`50
`
`100
`
`150
`
`0
`
`50
`
`100
`
`150
`
`0
`
`50
`
`100
`
`150
`
`0
`
`50
`
`100
`
`150
`
`100
`
`50
`
`0
`
`150
`
`100
`
`50
`
`0
`
`0
`
`50
`
`100
`
`150
`
`0
`
`50
`
`100
`
`150
`
`50
`
`100
`
`150
`
`50
`
`100
`
`150
`
`Figure 6. Experiment result of autonomous hovering
`
`A series of experiments has been performed using the
`proposed
`controller on Yamaha R-50. The
`experiment is performed as described here: The
`RUAV is first placed at the test field, preferably flat
`
`location, and then the flight system and radio/servo
`system are turned on. The GPS system automatically
`starts tracking the GPS satellite signals to compute
`the current location coordinates. The DGPS broadcast
`are used to achieve the ultimate accuracy of 2cm.
`Once GPS is completely locked on, the DQI-NP
`system is initialized and sequenced into the fine-
`alignment mode. During this process, the DQI system
`should be left undisturbed to correctly estimate the
`scale and bias factors of the six solid state inertial
`sensors. After these two navigation sensors are
`initialized, the engine is started manually and the
`RPM is quickly brought up to 90 % of the hovering
`RPM
`to avoid any
`low frequency mechanical
`resonance harmful to proper INS/GPS operation.
`
`During the repeated experiments, the attitude/velocity
`controller has shown stable operation even when the
`helicopter stays on the ground. Therefore, more
`accurate take-off and landing can be achieved by
`activating the attitude/velocity controller even before
`the helicopter takes off from the ground. When
`operated manually,
`the
`pilot
`engages
`the
`attitude/velocity controller using a switch on the
`transmitter and then takes the helicopter off from the
`ground. At this time, only steady heave reference
`command is given. Once the helicopter reaches the
`desired altitude, the hovering controller, i.e., the
`position/velocity/attitude loop controller is activated.
`
`Figure 6 shows the experiment result of hovering
`controller tested on R-50 UAV. The RUAV showed a
`stable response over
`two minutes with ±0.5m
`accuracy
`in x and y direction. Roll, pitch,
`translational velocity in x and y directions are
`regulated very well altogether. The altitude regulation
`shows rather large variation because the engine was
`
`not regulated to a constant RPM. A simple
`proportional-integral(PI) controller will be added in
`system for more stable response.
`
`5. Vehicle Control Language Framework
`
`Once the regulation layer for the vehicle control is
`designed, then supervising control logic should be
`integrated with the vehicle regulation layer to guide
`the RUAV along the desired trajectory. This layer
`plays an important role to relay the mission control
`layer and
`the vehicle-level control
`layer by
`generating appropriate reference trajectories and then
`injecting them into the proper controller, which is
`activated accordingly depending on the flight status
`and the target mode.
`
`American Institute of Aeronautics and Astronautics
`
`Parrot 1021, Page 7
`
`

`
`(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
`
`Vehicle Control Language, or VCL, is an application
`and development tool for UAV systems, operating in
`a hierarchical structure as shown in Figure 7. This
`approach allows to program complex behavior of the
`UAV adaptively without rewriting program as the
`mission changes. The sequence of motion commands
`is described in a script language form understandable
`to human. VCL module consists of user interface part
`on ground station, language interpreter and sequencer
`on the UAV side.
`
`shown in Figure 9. The VCL execution module
`(VCLEM) selects the proper controller depending on
`reference
`the
`flight mode and generates
`the
`command. VCLEM monitors the vehicle trajectory
`and determines if one sequence is finished or not. It
`also monitors the vehicle status for possible troubles
`in sensor or the vehicle itself. If error detected, the
`fault detection algorithm shown in Figure 5 is
`activated and proper error handling measure is
`executed. In the worst case, the VCL releases the
`automatic vehicle control mode and return the control
`to the ground-based test pilot. This routine is repeated
`until the end of VCL command script is reached and
`the RUAV returns to its default flight mode. Finally,
`sample VCL codes are given in Figure 10.
`
`Figure 7. Hierarchical structure of VCL processing
`
`Output
`
`/
`
`______Oofarer.^ /
`
`/
`
`Figure 9. Flowchart of VCL execution module
`
`6. Conclusion
`
`of
`
`development
`the
`introduced
`This paper
`hierarchical RUAV controller design conducted at
`University of California at Berkeley. The vehicle
`dynamics model is identified using the parametric
`model suggested by Mettler. A different approach
`based on the time-domain analysis tool is applied to
`the experiment data and linear time domain model is
`obtained. The low-level vehicle control layer based
`classical SISO approach and implemented on a
`Berkeley RUAV, Yamaha R-50. The proposed
`controller showed satisfactory autonomous hover
`flight capability. The controller, or any other vehicle
`
`Figure 8. Waypoint generation using graphical user
`interface
`
`When a mission is given, the ground operator
`generates a sequence of waypoints with
`their
`attributes such as the type of waypoint, heading,
`velocity and so on as shown in Figure 8. When
`finished, the VCL is uploaded to the RUAV control
`system and then executed in sequential manner as
`
`American Institute of Aeronautics and Astronautics
`
`Parrot 1021, Page 8
`
`

`
`(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)1 Sponsoring Organization.
`
`level controller, is integrated with higher level
`behavioral control logic called VCL. Currently, the
`proposed paradigm
`is being
`implemented on
`Berkeley UAVs and will be tested in near future.
`
`[6] D. H. Shim, H. J. Kirn, S. Sastry, "Control System
`
`Design for Rotorcraft-based Unmanned Aerial
`Vehicles using Time-domain System Identification,"
`IEEE
`Conference
`on Control Applications,
`Anchorage, Alaska, 2000.
`
`GO AUTO
`WaitFor Go
`TakeoffTo(0,0,-2)abs for 30sec
`FlyTo (5,10, Orel vel Smps
`passby autoheading
`FlyTo (0,5,0)rel vel lOmps
`stopover autoheading
`Hover (0,0,0)rel for 20sec
`FlyTo (8,-5, Orel vel Smps
`passby autoheading
`FlyTo (-5,-10,OJrel vel Smps
`passby autoheading
`FlyTo (-5,0,0)rel vel Smps
`stopover autoheading
`FlyTo (0,0,0labs vel Smps
`passby autoheading
`Hover (0,0,0)rel for 40sec
`Land
`GO MANUAL
`
`[7] R. T. N. Chen, M. B. Tischler, "The Role of Modeling
`and Flight Testing in Rotorcraft Parameter Identification,"
`The 1986 AHS Forum, System Identification Session,
`Washington, D.C., 1986.
`
`[8] L. J. Ljung, Matlab System Identification Toolbox
`User's Guide , The Math Works, Inc.,1997.
`
`[9] L. J. Ljung and E. J. Ljung, System Identification:
`Theory for the User, 2nd ed., Prentice-Hall, 1998.
`
`Figure 10. Sample VCL code for free-style waypoints
`
`7. Acknowledgement
`
`This research was supported in part by the Army
`Research Office DAAH04-96-1-0341 and by the
`Office of Naval Research under Grants N00014-97-
`1-094.
`
`8. Reference
`
`[1] R.W. Prouty, Helicopter Performance, Stability
`and Control, Krieger Publishing Company, 1995.
`
`[2] M. B. Tischler, "System Identification of Small-
`Size Unmanned Helicopter Dynamics," American
`Helicopter Society 55th Forum, Montreal, Quebec,
`Canada, May 1999.
`
`[3] H. Shim, T. J. Koo, F. Hoffmann, S. Sastry, "A
`Comprehensive Study of Control Design for an
`
`Autonomous Helicopter," 37th
`IEEE Conference
`Decision and Control, pp. 3653-3658, 1998.
`
`on
`
`[4] C. P. Sanders, P. A. DeBitetto, E. Feron, H. F. Vuong,
`N. Leveson, "Hierarchical Control of Small Autonomous
`Helicopters," 37th IEEE Conference on Decision and
`Control, pp. 3629-3634, 1998.
`
`[5] J. E. Corban, A. J. Calise, J. V. R. Prasad,
`"Implementation of Adaptive Nonlinear Control for Flight
`Test on an Unmanned Helicopter," 37th IEEE Conference
`on Decision and Control, pp. Pp. 3641-3646, 1998.
`
`American Institute of Aeronautics and Astronautics
`
`Parrot 1021, Page 9