Developing Embedded Control Systems with XtratuM

Application to Control the Attitude of a Mini-helicopter

P. Garc´ıa

, P. Albertos

, A. Crespo

and F. Gonz´alez

Institute of Autom´atica e Inform´atica Industrial, Universitat Politecnica de Valencia, Valencia, Spain

Universidad Central de las Villas, Santa Clara, Cuba

Keywords:

Embedded Control Systems, Development Platform, UAV Control.

Abstract:

Developing embedded control systems requires to have the possibility of analyzing and evaluating new control

algorithms before their implementation in the ﬁnal system as well as their robust operation once implemented.

For that purpose, a generic platform composed by the hypervisor XtratuM and RTOS PaRTiKLe is introduced

in this paper. The interaction between the user and the process is generated by using Linux, while the real-

time execution of the process is ensured by PaRTiKLe. Moreover, hypervisor XtratuM provides the facilities

to execute several partitions with different guest OSs as well as the mechanisms to communicate PaRTiKLe

partition and the control environment. By means of this platform the performance of future partitioned em-

bedded systems are analyzed, evaluated and improved. Experimental tests are carried out in order to prove the

effectiveness of the system. The reported results show the good performance of the designed system and the

robustness of the platform.

1 INTRODUCTION

One of the latent problems when working on embed-

ded control systems (ECS) is the lack of a platform

to evaluate the control algorithm before being em-

bedded. This pre-validation of the control algorithm

would save substantial time, which can be normally

tedious, to adjust the controller gains. Moreover, po-

tential system failures can be detected before the con-

trol is put into operation and, in this way, avoid eco-

nomic losses.

Recent advances in software technology, have

produced a tendency to design, develop and inte-

grate more software and hardware components for

real-time embedded systems. Example can be found

in aeronautics, automobile equipment, in mobile

telecommunication, and so on. The capacity to sup-

port real time activities, device drivers access, fault

tolerance and minimal distribution of software and

hardware (Kopetz, 2011) are the main aspects to con-

ceive an ECS.

It was in the avionics ﬁeld were the partitioned

systems were proposed and developed in order to in-

crease the security and predictability of the control

systems. The foundations of this approach are pre-

sented in (Rushby, 1999). Today, partitioned systems

are being evaluated and promoted in several ﬁelds of-

fering the services and mechanisms to build up safe

and secure embedded systems (AUTOSAR, ), (M. H.

Deredempt, A. Crespo et al., 2012).

The use of partitioned systems permits to develop

mixed criticality applications where some partitions

may have different levels of temporal constraints or

security properties (Commision, 2012). In control

systems, the partitioned systems can involved one

or several partitions with real-time operating systems

executing control activities and other partitions with

general purpose OSs than can provide the graphical

and end-users interfaces and other remote services.

The system is executed on top of a hipervisor that pro-

vides virtual machines to the partitions.

This approach allows a feasible and optimized

software implementation with a better resource usage,

and this in turn results in a cost reduction. Therefore,

the main qualities of such a well-engineered software

should lie in its capacity to support fast, easy-to-use,

low-cost solutions, while guaranteeing, at the same

time, a predeﬁned quality of service. Moreover, eval-

uate the performance of the software and hardware is

also necessary and very important before the control

system is embedded.

Currently, there are evaluating methods for em-

bedded systems but they are only valid for speciﬁc ap-

plications or speciﬁc programming languages. Like-

434

García P., Albertos P., Crespo A. and González F..

Developing Embedded Control Systems with XtratuM - Application to Control the Attitude of a Mini-helicopter.

DOI: 10.5220/0004484504340441

In Proceedings of the 10th International Conference on Informatics in Control, Automation and Robotics (ICINCO-2013), pages 434-441

ISBN: 978-989-8565-70-9

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

wise, the use of these systems leads to high develop-

ment costs and maintenance, besides requiring a deep

knowledge of them. On the other hand, the imple-

mentation of control laws in these evaluation systems

must obey certain pre-established criteria for the sys-

tem. This can result in a problem, since the control

algorithm would have to be adjusted or reconﬁgured,

producing a change in its structure and in the system

behavior when ﬁnally applied.

To our knowledge, there is not yet a generic plat-

form, in process control, being able to implement dif-

ferent types of operating systems and/or processes.

The use of an open-source hypervisor and RTOS

(PaRTiKLe), could improve the development process

in many ways, such as a low price in the development

and maintenance and good robustness when other ap-

plications are running. Speciﬁcally, the model of de-

velopment conceived in the open source community

could be successfully adapted to the design and im-

plementation of real-time critical application, improv-

ing the reliability and the extensibility of the software

components.

In this paper we introduce a new partitioned plat-

form composed by XtratuM and PaRTiKLe systems,

allowing a robust design of embedded control sys-

tems. A Linux partition provides the advantage of the

interaction between the user and the process, while

PaRTiKLe partition ensures the real-time execution

of the control tasks even when the system is run-

ning other tasks. XtratuM allows for strong safety of

the application, always required in ECS. The major

applications of the platform are to analyze, evaluate

and improve the performance of future embedded sys-

tems. Additionally, the proposed platform can be used

in critical embedded systems, i.e. where the dynamics

of the system is very fast and the processing time must

be executed under strict constraints. A more detailed

description of the proposed platform is presented in

section 2.

The validation of the platform and its application

to the design of embedded control systems for fast

dynamic plants (UAV) are introduced in section 3. A

comparison of the proposed platform with respect to

a platform under only Linux RT environment is also

provided. Finally, some comments and conclusions

are included in the last section.

2 SYSTEM DESCRIPTION

Embedded control systems are increasingly being

used in safety critical applications and infrastructures

such as, aircraft ﬂight control and electric grids (Cre-

spo and Alonso, 2006) (Parkinson and Kinnan, 2003)

(J. Loyall and Fernandez., ). It is well known that,

the integrity of these critical systems will depend on

the good performance of the installed software com-

ponents. And, a bad performance or losing data in

the framework, may cause massive material and even

human life losses.

In many critical systems, and the proposed plat-

form is not an exception, it is required that numerous

software functions with varying levels of exigency

(for example, hard, soft and non real time tasks) are

running at the same time. This kind of integration

creates a special challenge because it is required to

guarantee the execution of every task and the non in-

terference with each other. In (Masmano et al., 2009),

the authors present a software architecture based on

a virtualization layer offering virtual machines to ex-

ecute partitions. This approach offers the capability

to deﬁne several partitions guaranteing their temporal

and spatial isolation and achieving the deterministic

behavior required by safety-critical systems.

To ensure the previous requirements, the designed

platform uses an open source based hypervisor (Xtra-

tuM) which executes real-time constrained partitions

based on PaRTiKLe RTOS. Moreover, this fact will

facilitate the reuse of the code and will enable the de-

velopment of more complex software without sacri-

ﬁcing real-time performance. In the following, the

main characteristics of XtratuM and ParTiKle sys-

tems are described.

2.1 XtratuM

XtratuM is an hypervisor for embedded devices,

it provides a framework to run multiple concur-

rent operating systems in a robust partitioned en-

vironment (Masmano et al., 2005). XtratuM can

be also deﬁned as an open source nanokernel

(www.opensource.org/docs/deﬁnition.php, ), just im-

plementing two simple device drivers: interrupt and

timer. Likewise, XtratuM can be considered as a

small subset of the lowest operating system layers that

can meet the hard real-time system requirements: fast,

compact, portable, simple and predictable.

The main characteristics of the XtratuM system

are (see Figure 1):

• Bare metal hypervisor designed for real-time em-

bedded systems.

• Several partitions can be simultaneously executed

• Every partition contains the most appropriated OS

according to the application needs.

• Every partition is executed in separated memory

areas or regions. It prevents for errors or attacks

DevelopingEmbeddedControlSystemswithXtratuM-ApplicationtoControltheAttitudeofaMini-helicopter

435

outside of the partition creating a new and higher

level of security.

• Every partition may have different level of secu-

rity.

Figure 1: XtratuM structure.

For critical control processes, these features are

very important, because the execution of two different

systems at the same time, should give quick access to

some devices or ﬁles and an easy interaction between

the user and the process, and at the same time guaran-

tee the implementation and execution of the tasks in a

RTOS domain. Additionally, the last two feature are

essential to fulﬁll security requirements in security-

critical real-time systems. For example, working in

a system composed by Linux and RTOS domains, if

a computer attack happens to the Linux domain, it

should not affect the control law execution embedded

in the RTOS domain.

An example of this can be an engine control pro-

gram with a graphic interface. The application can

be split into two parts: the control algorithm which

interacts with the process to be controlled, with hard

real-time and fault-tolerance requirements (where the

use of a hard RTOS is compulsory) and the graphic

interface, without real-time requirements, displaying

interesting data or accepting some input commands.

Figure 2 introducesGNU/Linux and PaRTiKle en-

vironments implemented in XtratuM (see details in

section 3.2). The main characteristic of this conﬁgu-

ration is the effectiveness to separate the priority tasks

being executed by ParTiKle from the other ones per-

formed by GNU/Linux. This fact yields the system’s

behavior stable even if an error appears in the general

system. The PaRTiKle partition forward data to the

Linux one, to be independently treated, and receives

from that one updating of the control algorithms or

tuning parameters.

MMIandother

applications

LINUX

RT-tasks

PaRTiKle

Criticalprocess

UFO

Internal

Communication

Channels

Wireless

Figure 2: Critical process and software platform.

2.2 PaRTiKle Partition

PaRTiKle is an embedded real-time operating system,

distributed under the terms of the GNU Public Li-

cense (Peiro et al., 2007). It provides features such as

full preemptability, minimal interrupt latencies, syn-

chronization primitives, scheduling policies, and in-

terrupt handling mechanisms, needed for critical ap-

plications with hard real time constrains.

Additionally, PaRTiKle brings support, in a bare

machine, for multiple execution environments such as

for example, a Linux regular process and a hypervi-

sor domain, see Figure 3. Although, PaRTiKle was

designed to be POSIX compatible, it can also provide

support forC++, Ada and Java languages, and it fur-

nishes a standard C library in the application context.

Hardware

XtratuM

PartiKle

Control tasks

Linux

Applications

Realtime

partition

Nonrealtime

partition

Figure 3: PaRTiKle as XtratuM domain.

ICINCO2013-10thInternationalConferenceonInformaticsinControl,AutomationandRobotics

436

These characteristics help to design different levels of

abstraction in the process and control domains. This

is very useful to reuse the code and to make shorter

the development cycle.

Likewise, the PaRTiKle’s kernel provides addi-

tional characteristics like thread scheduling, synchro-

nization, timer and communication primitives. On the

other hand, it also handles hardware resources, such

as, interrupts, exceptions, memory, timers, etc.

A brief summary of the main characteristics of

PaRTiKle system as it is implemented nowadays is

as follows:

Scheduling: The scheduler only implements a Rate-

Monotonic (RM) scheduling policy.

Timing Management: It provides multiple virtual

timers to the kernel and the application.

Physical Memory Management: It is able to man-

age the free available physical memory.

Kernel Library: There is only a minimal C library

used by the kernel. This library cannot be ac-

cessed by the application.

System Call: All kernel services are provided via a

single entry point.

2.3 The Designed Platform

The physic platform is composed by a processor (up

or equal to pentium II), the software described previ-

ously, and the input/ouput cards. The platform can be

thus connected to the process to control or evaluate.

The items added to the platform will depend to the

process to study.

The designed platform presents some advantages

with respect to others in the literature, such as

• Some tests, developed by using several Linux

tools, can be done using PaRTiKle, like a Linux

process.

• The platform can be executed using PaRTiKle like

being a XtratuM domain and, at the same time,

take advance of the Linux domain (persisting data,

monitoring some value, etc).

• The ﬁnal program can be embedded using PaR-

TiKle as a bare machine.

• The platform is portable (different hardware un-

der the same software can be implemented), con-

ﬁgurable and maintainable (Peiro et al., 2007).

In addition, all the main components in the plat-

form, GNU/Linux, PaRTikle and XtratuM, are based

on the well known open standard languages (POSIX,

ARINC)

. Likewise, future works in the proposed

platform will use some important advantages of these

languages, for example, the temporal nature of Ada

or the facilities to create strong interfaces in C++ and

Java.

To prove the robustness of the proposed platform

and to verify the behavior of some important param-

eters, a real-time analysis of the experiment can be

done using the jitter performance evaluation tool (as

reported in (Berna et al., 2011)). In fact, it is well

known that certain values of the jitter can degrade the

control performance and in extreme cases even cause

instability of the closed-loop system, see (Stothert and

MacLeod, 1998) (Shin and Cui, 1995).

In the experiments here performed, heavy compu-

tation load tasks have been concurrently run with the

RT control tasks. If the load is too high, there are not

enough resources. The control tasks performance is

shown not being affected by the system degradation.

3 PLATFORM VALIDATION

To validated the proposed platform two kinds of

experiments are presented in this section. First,

a Quanser Helicopter with four rotors is used

((Quanser, 2007)). The performance of the platform

is tested in order to stabilize the roll angle in a desired

position. This is a ’subcritic’ process because the

helicopter is placed on a base platform in such a

way that the attitude is the only possible variable

to be stabilized. This base platform avoids the

helicopter crashing in case that the controller is not

able to stabilize it. The second experiment is also

devoted to stabilize the attitude of the helicopter

but ﬂying in 3D. In this helicopter any wrong

measurement from the sensors and/or malfunction

of the microcontroller will result in a helicopter crash.

The two processes are tested using the same plat-

form and they can be selected by using a switch, see

Figure 4.

Both processes are helicopters and the goal is to

stabilize the attitude dynamics for each one. There-

fore, the classical attitude mathematical model is

given by (Castillo et al., 2005)

η = τ

∀ η = ψ, θ, φ (1)

where η represents the Euler angles and τ

the control

input.

Under the POSIX standard design guides there are a lot

of good products, several Unix versions are among the most

successful ones. The ARINC speciﬁcation was designed to

be used by high-reliable products and is used in avionics

applications.

DevelopingEmbeddedControlSystemswithXtratuM-ApplicationtoControltheAttitudeofaMini-helicopter

437

Figure 4: Platform & process schema.

3.1 Application to the Quanser

Helicopter

In this section, the proposed platform is tested

stabilizing the roll angle (φ) of a mini helicopter

with four rotors mounted in a vertical base (see more

characteristic of the helicopter in (Quanser, 2007)).

The helicopter is a modiﬁed Quanser helicopter

connected with the processor using the input/output

cards. Moreover, the helicopter can be only con-

trolled in attitude. An IMU (MicroStrain, 2007)) is

used to measure the orientation (ψ, θ, φ) and the angu-

lar rate (

ψ,

θ,

φ) of the vehicle. The IMU is connected

via an USB port. In addition, the system receives

commands from a small keyboard and will send, peri-

odically, the system status to a host computer in order

to display and analyze the system variables and status.

The following controller, based on saturation

functions, is used to stabilize the roll angle of the he-

licopter

= −σ





− σ



(φ− φ

)



(2)

The stability analysis of the closed-loop system is

given in (Sanahuja et al., 2010).

The behavior of the platform described in section

2 is compared with respect to a platform only based

on the Linux RT system. The control objective is that

the helicopter should follow a step trajectory. Once

this task is completed, more tasks are added to the

processor in order to saturate its capacity and ver-

ify the system’s robustness. Figure 5 shows the φ

response when applying the previous control law in

Linux RT system. Solid line represents the almost

perfect system response without others tasks execut-

ing, while the dotted line represents the same response

when others tasks need to be executed (stress condi-

tions). Note from this ﬁgure that, when executing the

control algorithm at the same time with others tasks,

and increasing the load, the system becomes unstable

for some given load. The control inputs generated in

these experiments are shown in Figure 6.

The φ behavior when the proposed platform (Xtra-

tuM + PartiKle system) is used by applying the con-

0 1000 2000 3000 4000 5000 6000 7000 8000 9000

−10

−5

Time(seg)

φ(º)

Figure 5: φ Angle response.

0 1000 2000 3000 4000 5000 6000 7000 8000

−15

−10

−5

Time (s)

(V)

Figure 6: τ

Control input.

trol law (2) to stabilize the roll angle, is shown in Fig-

ure 7. This experiment is also done concurrently with

(or without) others tasks. The solid line represents the

system response without others tasks while the dotted

line represents the same response with others tasks

executing at the same time. Note from this ﬁgure that

the system remains stable in both cases and, there-

fore, these experiments illustrate the robustness of the

platform.

3.2 Application to the XUFO

The XUFO platform is more complete and it is

composed by a processor, an IMU, a radio, a heli-

copter and a joystick, see Figure 8. The employed

helicopter is a mini-helicopter with four rotors

from X-UFO company (Xufo, 2002)). It is a well

known helicopter used in some control aeronautical

applications. The main characteristics are that the

front and the rear motors rotate clockwise while the

other two rotate counter clockwise, the gyroscopic

phenomena are relatively small. This helicopter does

not have a swatch plate. In fact it does not need any

servomechanism. The main thrust is the sum of the

ICINCO2013-10thInternationalConferenceonInformaticsinControl,AutomationandRobotics

438

0 500 1000 1500 2000 2500 3000 3500 4000 4500

−8

−6

−4

−2

Time (s)

φ (°)

Figure 7: φ Angle response.

Figure 8: Control platform schema.

thrusts of each motor. Pitch movement is obtained by

increasing/reducing the speed of the rear motor while

reducing/increasing the speed of the front motor.

The roll movement is obtained similarly using the

lateral motors. The yaw movement is obtained by

increasing/decreasing the speed of the front and rear

motors while decreasing/increasing the speed of the

lateral motors. This should be done while keeping the

total thrust constant. The helicopter evolves freely in

a 3D space without any ﬂying stand.

The radio is a Futaba Skysport 6. The radio and

the PC are connected using data acquisition cards.

The connection in the radio is directly made to the

joystick potentiometers for the trust, yaw, pitch and

roll controls. To measure the orientation and the an-

gular rate of the vehicle, a wireless IMU (Micros-

train, (MicroStrain, 2007)) is used. The IMU sends

this information via wireless to a receiver located in

the ground station (processor). Moreover, the com-

puted control inputs are sent to the rotors via a dig-

ital/analogic converter. Additionally, the system can

receive commands from the keyboard and send, peri-

odically, the system status to a host computer in order

to display and analyze the system variables.

Experimental Results

Several experiments have been done. To illustrate the

results, the following control is applied in order to sta-

bilize the yaw angle under aggressive disturbances.

The following nonlinear control law based on satura-

tion functions has been used to control the system (1),

= −σ

η) − σ

(η− η

)) ∀ η = ψ, θ, φ

(3)

where |σ

(◦)| ≤ η

, ∀ i = a, b, is a saturation func-

tion, a, b > 0 are constant and k

> 0, ∀ j = 1, 2, are

constant. Introducing (3) into (1), we obtain

η = −σ

η) − σ

(η− η

)) (4)

The stability analysis of the system (4) was also pre-

sented in (Sanahuja et al., 2009). The main character-

istic of this control strategy is that when the system is

working in the linear part of the saturation function it

can be seen as a linear controller.

Yaw Stabilization. The idea here, is to apply the

control law in open-loop and closed-loop in order to

analyze the performance of the platform. Addition-

ally, some manual aggressive perturbations and a de-

sired trajectory are added on-line to validate the ro-

bustness of the system.

The performance of the controller used to stabi-

lize the yaw angle of a helicopter is shown in Fig-

ure 9. Some forces have been manually (or using

the joystick) added to perturb the system. Note that,

the closed-loop system remains stable in the presence

of aggressive perturbations. To improve the perfor-

mance of the experiments the controller parameters

have been tuned on line. On the other hand, a stress

condition is given to the processor, executing a video

at the same time, in order to perturb the process. In

pure Linux environmentsthe use or execution of other

applications could “stop” the process, and notice that

it is not the case for the developed platform. Figure

10 shows the control input applied to the system.

4 PARTITIONED APPROACH

ANALYSIS

When comparing a partitioned approach with respect

to a more classical approach, the following advan-

tages based on the temporal and spatial isolation to

partitions provided by the hypervisor can be men-

tioned:

• Separation of concerns: Each partition can use the

most approriated operating system. In this case,

DevelopingEmbeddedControlSystemswithXtratuM-ApplicationtoControltheAttitudeofaMini-helicopter

439

0 5000 10000 15000

−200

−150

−100

−50

100

150

200

Time (s)

ψ (°)

Ref.

yaw

Perturbations

closed−loop system

Open loop system

Figure 9: ψ Angle response.

0 5000 10000 15000

−0.4

−0.2

0.2

0.4

0.6

0.8

1.2

Time (s)

(V)

Figure 10: τ

Control input.

real-time tasks require a small real-time system.

On the other hand, the set of activities related with

graphical interface, network communications, re-

mote access, etc., may require a fulloperating sys-

tem as Linux. It allows the integration of real-time

with non real-time activities in the same platform.

• Several levels of criticality: each partition may

have different real-time constraints and, conse-

quently, different level of validation, security, etc.

• Independent development: partitions can be inde-

pendently developed and validated. Very well de-

ﬁned mechanisms to communicate partitions are

provided by the hypervisor.

• Faults are limited to the faulting partition: parti-

tioning approach allows to have independent par-

titions that are not affected by the faults in other

partitions. The hypervisor grants that a fault in

a partition is not propagated to other partitions.

In this case, the real-time partition deﬁnes a lim-

ited set of tasks that can be fully validated jointly

with the real-time operating system using the ap-

propriated techniques to achieve a secure parti-

tion. On the Linux partition, it is more complex

(and costly) to fully validate the interfaces, com-

munications, etc. If a fault occurs in the Linux

partition, the hypervisor conﬁnes the fault to that

partition without impacting the execution of other

partitions.

• External attacks: if the Linux partitions permit the

external access to the application services, some

external attacks may happen. Even if a hacker

gets the Linux partition control, the spatial par-

tition isolation grants that no access to other par-

titions is allowed.

However, some drawbacks can be also consider-

ated:

• More complex software architecture: the software

architecture is more complex due to the existence

of a hypervisor and several guest OS for the parti-

tions.

• Scheduling: a cyclic scheduling has to be de-

signed to fulﬁl the temporal requirements of the

real-time partitions.

• System deployment: the ﬁnal system to be de-

ployed in the embedded hardware platform has to

include the hypervisor and the partitions.

5 CONCLUSIONS

A new test-bed platform has been presented in order

to validate and improve the design of embedded con-

trol systems. To run different tasks in an independent

way, providing maximum operation safety, a hipervi-

sor (XtratuM) has been used at the lower level.

Linux RT and Partikle like O.S. have been used in

order to implement the control strategies and to guar-

antee priority tasks.

The proposed platform has been used for validat-

ing the ECS designed for a quadrotor and a lab heli-

copter. It has been proved to be robust under different

load conditions.

The platform is versatile and allows the use of dif-

ferent OSs as well as control algorithms, with the op-

tion to experiment with changes in the control laws,

control parameters and operating conditions.

Once the control structure is validated, it can be

deployed and implemented in the ﬁnal system.

REFERENCES

AUTOSAR. Internet Automotive open System Arquitecture.

Berna, A., Castillo, P., Sanahuja, G., Gonz´alez, F., Garcıa,

P., and Albertos, P. (2011). Development of a test-bed

ICINCO2013-10thInternationalConferenceonInformaticsinControl,AutomationandRobotics

440

to implement and validate real-time control strategies

for aerial vehicles. In Proceedings of the 18th IFAC

World Congress.

Castillo, P., Lozano, R., and Dzul, A. E. (2005). Modelling

and control of mini-ﬂying machines. Springer.

Commision, E. (2012). Workshop on mixed criticality sys-

tems. http://cordis.europa.eu/fp7/ict/computing/home

en.html.

Crespo, A. and Alonso, A. (2006). Una panor´amica de los

sistemas de tiempo real. Revista Iberoamericana de

Autom´atica e Inform´atica Industrial (RIAI), 3(2):7–

18.

J. Loyall, R. Schantz, D. C. and Fernandez., S. A dis-

tributed real-time embedded application for surveil-

lance, detection, and tracking of time critical targets.

11th IEEE Real-Time Embedded Technology and Ap-

plications Symposium,11th IEEE Real-Time Embed-

ded Technology and Applications Symposium, 2005.

Kopetz, H. (2011). Real-time systems: design principles

for distributed embedded applications, volume 25.

Springer.

M. H. Deredempt, A. Crespo et al. (2012). Integrated

modular avionics for spacecraft software architecture

and requirements. In DASIA 2012. DAta Systems In

Aerospace.

Masmano, M., Ripoll, I., and Crespo, A. (2005). An

overview of the xtratum nanokernel. In Proceedings

of the Workshop on Operating Systems Platforms for

Embedded Real-Time Applications (OSPERT).

Masmano, M., Ripoll, I., Crespo, A., Metge, J., and Ar-

beret, P. (2009). Xtratum: An open source hypervisor

for TSP embedded systems in aerospace. In DASIA

2009. DAta Systems In Aerospace.

MicroStrain (2007). 3DM-GX2TM Data Communication

Protocol, technical report version 1.13 edition.

Parkinson, P. and Kinnan, L. (2003). Safety-critical soft-

ware development for integrated modular avionics.

Embedded System Engineering, 11(7):40–41.

Peiro, S., Masmano, M., Ripoll, I., and Crespo, A. (2007).

Partikle os, a replacement for the core of rtlinux-gpl.

In Proceedings of the 9th Real-Time Linux Workshop,

Linz, Austria, page 6.

Rushby, J. (1999). Partitioning in avionics architectures:

Requirements, mechanisms, and assurance.

Sanahuja, G., Castillo, P., and Sanchez, A. (2009). Stabi-

lization of n integrators in cascade with bounded input

with experimental application to a VTOL laboratory

system. International Journal of Robust and Nonlin-

ear Control. DOI: 10.1002/rnc.1494.

Sanahuja, G., Castillo, P., and Sanchez, A. (2010). Stabi-

lization of n integrators in cascade with bounded input

with experimental application to a vtol laboratory sys-

tem. International Journal of Robust and Nonlinear

Control, 20(10):1129–1139.

Shin, K. G. and Cui, X. (1995). Computing time delay

and its effects on real-time control systems. Control

Systems Technology, IEEE Transactions on, 3(2):218–

224.

Stothert, A. and MacLeod, I. (1998). Effect of timing jitter

on distributed computer control system performance.

In Proceedings of 15th IFAC Workshop DCCS, pages

25–30.

Quanser, I. (2007). “http://www.quanser.com/”.

Xufo, I. (2002). “http://www.xufo-shop.de”.

www.opensource.org/docs/deﬁnition.php. Open source ini-

tiative, open source deﬁnition. published online in

open source initiative’s ofﬁcial web site. December

20095.

DevelopingEmbeddedControlSystemswithXtratuM-ApplicationtoControltheAttitudeofaMini-helicopter

441