You are on page 1of 10

Embedded Systems in Real Time Applications, Design & Architecture


Motivation Necessit y is the moth er of invention and embedded system s are inventions that were fuelled by the idea of making pre-program s to perform a dedicat ed narrow range of funct ions as part of large sy stem s. Usual ly with minimal end user int eractions, the 'giant leap tech nology' in future embedded system s is based on instr uction-orient ed design but not on design-oriented in structi ons. Embedded syst ems and real time operating syst ems (RTOS) are fast achi evi ng ubiquity, blurring the lines between sci ence fiction and hard reality. In general an RTOS h as the following futures: 1. multitaski n; 2. process threads that can be prioritized; 3.A suffici ent number of interrupt levels. An embedded syst em is any device controlled by instructions st ored on a chip. These devices are u sually controlled by a microprocessor that execut es the instructions stored on a ROM chi p. Embedded system s are used in navigati on tools like global positioni ng sy stem (GPS), automated teller machines (ATMs), networking eq uipment, digital video cameras, mobile phones, aerospace applications, telecom applicatio ns, etc. We concern ourselves wit h the development and implementation of model-based, realtime, embedded, hybrid control software. In particular, we target intelligent cruise control appli cations, including Adapti ve Crui se Control (ACC), in which a forward-loo king range sen sor (radar or Lidar, usually) is used to follow a vehi cle, and Coo perati ve ACC (CACC), a variation in which wirel ess communi cations are used t o supplem ent the forward looking sensor. We discu ss modeling on automated v ehicl es. Our approach emphasizes the mai ntenance of a si ngle m odel throughout the devel opment process, with particular emphasi s on "tigh t-loop."

Embedded syst ems and Real Time Operating system s (RTOS) are two among the several technol ogies that will p lay a major role in making these concepts po ssible. A large numb er of people are already depending on operating syst ems for real time applicatio ns, these 'eyes in the sky' are now going to make an impact on our every day lives in a more significant manner. What kind of help wil l these 'embedded syst ems' render unto humankind in th e future? Even Nostradam us would have been hard

pressed to an swer thi s question. Embedded syst ems ar e pre-designed without connections an d operate as per th e required task. But in operating syst ems instructi on is design-ori ented. These syst ems are basicall y platform-less sy stem s. Embedded system s are the unsung heroes of much of the technology we use today -- the video game we play, or the CD player or the washing machines we use em ploy them. Without an embedd ed sy stem we would not ev en be able to go online using modem. Design O rientation Embedded sy stem s are u sually low cost and are easily available off the shelf for most ap plications. They usually have low design risks, since it is easy to verify the design usi ng tool s fueling the growth of embedded syst ems. Embedded system s have received a major shot in the arm as the result of three develop ment s: The first was the d evelopment of standard run-time platforms like java, which enabl ed their use in myriad ways that were unimaginable in the past. The second was the coming together of embedded system s and the Internet, which made possibl e the net working of sev eral embedded system s to operat e as p art of a large system across networks. The third was the em ergence of sev eral integrated software environment s, which simplified the implementation s of these application s.

During operation, the design struct ure may be changed as per our task s. For example, con sider two transi stors; we can mould them using oth er passive el ement s as emitter coupled circui t, Darlington pair, etc., as per instruction. Real Time Appli cations Automobil es : Almost every car that roll s off the production line these days makes use of embedded technology in one form or the other; most of the embedd ed sy stem s in automobiles are rugged in nature, as most of these syst ems are made up of a single chip. No driver clashes or 'system s busy ' conditions happ en in these syst ems. Their compact profil es enable them to fit easily under the cramped hood of a car . These system s can be used to implement features ranging from adjustment of the suspension to suit road conditions and th e octan e content in the fuel to antilock braking systems (ABS) and security sy stem s.






Embedded syst ems can also make drive-l ess vehicle control a reality. Major autom obile manufactur ers are already engaged in work on these concepts. One such technolo gy is Adaptive Cruise control (ACC) from Ford. ACC allows car s to keep safe di stances from other vehi cles on busy highways. The driver can set the speed of his car and the dist ance between his car and others. He can over side the system anyt ime he wants by braking. Each car with ACC has a microwave radar unit or laser transceiver s fixed in front of it to determine the dist ance and relativ e speed of any vehicl e in its pat h. The ACC computer constantl y control s the throttle and brakes of the car. Another revolution is the way Internet servi ces will be integrated into the car. So when you drive past your mech anic's, you will be reminded that that your engine oil needs a refill, and when you cross the city limits, the toll will automati cally get deduct ed from your bank accou nt. And while passing the shopping mail, your PDA, which is connected to the Net via the car, will inform you about a new scale. In fact, the automati c to;l deduction con cept is already in effect in several countries around the gl obe.

Hybrid verificati on of the controller and analysi s of timing properti es are conduct ed through the use of third party t ools .

The approach i s appli ed to Adaptive Crui se Control (ACC) and Cooper ative ACC syst ems. Whil e regular cruise control system s track a desired vehicl e speed, Adapti ve Crui se Co ntrol (ACC) system s ad apt their behavior if there i s a vehicl e ahead on the roadway, and follow the leader vehicle at a driver requested tim e gap usi ng line-of-sight sensors such as radar and/or Lidar. When there i s no "leader" vehicle present, ACC default s to conventional crui se control and reverts to the driver- set speed. ACC syst ems ar e now avai lable on sev eral production car s, including the Nissan Q45 and FX45, the Mer cedes S-class, the Lexus 330 and 430, the Audi A8, and sel ect Jagu ar and Cadil lac models. These product ion ACC syst ems obtain their di stance and cl osing rate information about the leading vehicle throu gh the use of their forward-looking sen sor. These sen sors ar e typicall y subj ect to noi se, interference, false alarm s and dropouts, and th eir use requir es heavy filt ering. This in turn introduces delay s into the sy st em, and limits the abil ity of the ACC-equipped vehicles to follow the l eader vehicle cl osely or respond quickly to change in its speed. A variant of this i s Coop erati ve ACC (CACC), where the forward-looking sensor is compl ement ed by a wireless comm unication link that provides hop-by-hop, leader-to-f ollower updat es of critical information. Such a system can be designed to follow vehicl es wit h higher accuracy and faster respon se than traditi o nal ACC sy stem s, and should allow for freeway throughput cap acity increases. In addition, the CACC syst em can b e desi gned to have proven string stabilit y, so it could contribute to dam penin g shock waves in t he freeway traffic stream. Vehicle Model And Controll er Design The vehicle mod el used for controller development i s an eleven-st ate model, which incl udes vehicle st ate dynamics, throttle and brake system dynamics, a two-st ate model for the spark-ignition engine including external data maps which require interpol ation, and model s of the torque converter, tran smission an d wheel slip, as shown in the figure.

The vehicle state dy namics hav e two conti nuous states, vehicl e posi tion and veloci ty, and con sider vehi cle mass, air drag and rolling resi stance. The throttle and brake dy namics are both first-order, with one continu ous state for each representin g actuator dynam ics for the throttle and tim e respon se lag for the brakes. The controll er design process stem s from syst em requirem ents. Vehicles may be h et erogeneous, that i s of different types, mak es and model s. The controller was split hierarchicall y between an upper level controller that has several modes, nam ely cruise control (CC), adaptive cruise control (ACC) and coordinated adaptiv e cruise control (CACC). In ACC mode we use onl y information from the host vehicle's forward-lo oking sensors, and in CACC mode we suppl ement thi s information with dat a from the wireless communicati on sy stem.

The upper control generates a desired host vehicle accel eration, which is sent to the lower-level controller. The lower-level controll er convert s thi s desired acceleratio n to a desir ed torque, then chooses whether to apply the brakes or throttle, and in wh at amount. Both controllers are run on separ ate contro l computer s. In the following equation s, the following variables are u sed: Fa is the aerodynam ic drag force

Mrr is the rolling resist ance moment Rg is the gear ratio (related to engine and Vehicle speeds) ades is the desired syntheti c acceleration ct is the control torque h is the effective wheel radi us

Throttle control is used if:

For throttle control, the desired torque i s computed as:

Brake control is used if:

For brake control, the desired torque i s co mputed as:

1) Throttle control From the desired torqu e, the desired thrott le angle i s computed using an engine map. 2) Brake control From the desired torqu e, two different brake control strategies have been implement ed. In the first strateg y, the master cylinder pressur e is controlled. A pressur e regulator val ve con trols the pressur e applied on the hydraulic actuator. Seal friction exist s in the mast er cylinder and the actuator, and a small amount of hyster esi s is present in the pressure regulation v alve. The friction is modeled as hyperbolas from various points in the hyster esi s loop and can be written as Pmc = g (u) Feed-forward plus proportional feedback control is used, as develop ed. The control law can be written as:

Where ub is the applied command input to the brake sol enoid val ve, Pmc_des the desired master cylin der pressure, Pmc the measur ed master cylinder pressure, and kb>0 a feedback gain. In the second brak e control strategy, the wheel brak e pressure i s contr olled, and the brake system is modeled. The control law uses dynamic surface control and can be written as:

Where Pw_des i s the desired wheel pressu re, V is the volume of displ aced brake fluid, Pw the pressure at the wheel, Cq a flow coeffici ent, Cruise Control Law: The purpose of crui se control i s to maintain a desired velocity. A vehi cle may be in cruise control mode if it is not equipped with ACC or CACC, has no vehicle immediately in front of it or has at least 100 meter s of clearan ce to the preceding vehicl e, or by deci sion of the human driv er. The controller uses a feedback and feed-f orward control law of the form:

Where: a d is the desir ed accelerati on of the vehicl e, v is the speed of the vehicle, v d is the desired speed of the vehi cle and k is a gain set to 0.75. ACC and CACC Law: The control law for ACC and CACC i s identical. The main difference between both modes is in the sensor fusio n, and in the quality of the st ate information. Also, the operating logi c is d ifferent in both modes. The purpose of an ACC or CACC law is to regulate the range between veh icles to a user-select ed value, and to adju st the vehicle speed to the speed of downstr eam traffic. Velocity dependent (h eadway) control is used, with: Typical values for headway time range form 1.8 seconds to 0.7 seconds. The control law was designed using sliding control, where a surface is usuall y defined as a function of the error, derivatives of the error and/or integral s of the error. The surface is defined such that the st ate will exponentially decay al ong the surface to t he desir ed point. The input is chosen to guarant ee that the stat e will con verge and stay on the sliding surface. Error e is defined as: e = R Rd, where R=x1-x2. The sliding surface control is derived in two different ways, which basi cally lead to the same control law. Feedback linearization: Double first ord er method: Once again, the first two term s in the cont rol law are feed forward, and the last two ar e feedback. Both control laws are in essence equival ent.

Software Development Process A model-based approach is used throughout the control softwar e developm ent. Switchin g conditio ns from one mode to the next (for example ACC int o CACC) were designed by hand. The chosen architecture can th en be sim ulated, and C or C++ code can be generat ed for each task independentl y. This allows maintainin g a singl e model containi ng all of control and software information. The code that i s generat ed for the controller' s interfaces with legacy code, such as d evice drivers, driver displ ay unit s etc, through t he use of a shared-mem ory database on the "publi sh-and-su bscri be" model. On the experim ental test vehicles, all of the software is run on Pent ium computer s running the QNX4.25 real-tim e operating system.

Experimental Platform:

They are equipped with throttle, brake and steering actuati ng sy stem s, as well as wit h numerou s sensor s, including accel eromet ers, wheel speed sen sors, engi ne speed and manifold pressure sensors, as well as magnetom eter s that are used as part of the lateral control. In addition, both radars and the Lidar described above were mounted to the front bumper of the vehicl es. There are two con trol computers located in the trunk. Both run the QNX 4.25 operat ing system and communi cate ov er serial port con necti ons. The computers run a host of tasks necessary for automat ed control of the vehi cles, includi ng reading sensor data and writing to actuat ors, control com putations such as those described above for the ACC/ CACC system and low-level controller s, and tasks pert aining to driver display inform ation. There are about 30 different tasks runnin g on the most heavily loaded of the control computer s, and timing is fairly critical as human test driver s are in the cars during runs and their safet y is paramount. Other Applications Wired Wearables

A mobile phone in the form of a ring or earring? What ab out cool sungl asses, with streaming video di spl ays built into them? All these can soon be a reality. Embedded syst ems h ave a small footprint and consum e very little power, which makes them ideal for wearable comput ing applications. The minimal system requirements of the dev ices ensure that the hardwar e is almost microscopi c. IBM is already working on the prototype of a mobile phone th at can be worn as jewelry. The components of the phone will be di stributed among different pieces of jewelry -- earring, necklace, ring and bracelet. The phone is likely to have blue tooth capabil ity built into it. The earring will have embedd ed speak ers and will act as the recei ver. The necklace will have emb edded mi crophones that will act as mouthpiece user s can talk int o. IBM call s the ring par t of the phone the 'decoder ring'. Light emitting diodes (LED's) will flash to indicate an incoming call. The ring will also hav e features that will enabl e it to be programmed to flash different colour s for a particular user or to indicate the importance of a call. A video graphi cs array (VGA) will be built into the bracelet, which will displ ay the name and plans to incorporate voice recognition technology for dialing a number. The phone may also have features to indicat e new E-mail. Pacemakers Imagine a time when body tran spl ants like cardiac pacemak ers will be able to monitor & manage them selves rem otely. These syst ems wi ll be so compact that the pati ent would n't even be aware that they are embedded in his body, and developm ents are pointing t owards the use of pacemakers that can be tran spl anted in or near the heart itself. The pacemaker will be able to monitor param eter like bloo d pleasure blood flow, pressure rate temperatur e, etc., using microsensors placed in various parts of the body. The capabilit y will enab le the pacemaker to automatically vary it s operation to suit the changing body condit ions. It will also transmit dat a using micro sen sors pl anted in various par ts of the body. This capability will enabl e the pacemak er to automati call y vary its operati on to suit the changing body condi tion s. It will also transmit dat a using wir eless transmi ssi on. Embedded technology ad van ces l ess transmi ssion is likely to be done by a transmitt er implanted n ear the surface of the skin. In case in an abnormality i s detected. The doctor will be able to take rem edial action even from remote location s. A variety of operating system s are available for use wit h embedd ed computer s. Many of them ar e not true real -time operating syst ems (RTOS), as they do not sup port the preci se scheduling of tasks and predictab le react ions tim es of real-tim e events. A true RTOS must suspect prioritized, pre-emptiv e scheduling. Embedded computer s perform their jobs by executing software instructions. Unlike desktop computers, the user has little or no

information on what i s happening -- code is executed automati cally in respon se to 'real time' events. For exampl e, when an intruder opens a door connected to a security syst em, the microprocessor turns on an alarm s, dials the number of the security company, and transmit s an alarm signal. Smarter system s analyze dat a about the intrusion and turn on only if the intruder has hum an attribut es. Other than setting the alarm and checki ng for messages, the user of the alarm has no control over the software being execut ed. Conclusi on This paper presents the use of a model -base approach to the developm ent of real-time, embedded, hybr id control software. The concepts are illu strat ed with a scenario involving speed profile tracking and vehicl e following applications for usi ng the cruise controller. Rob otic technologies such as range, velocity and acceler ation measurement s, and their processing and fusion were used as part of the syst em. In addition, vehicles can present very nonlin ear behavior, especi ally at low speed s, and their control present s a formidabl e challenge. The problem domain of intelligent cruise control appli cations has been described in detail, along with control and software d evelo pment methodologies. All these application areas are ju st tiny drop s in the big ocean of embedded syst ems technology. These prov erbial Davids ar e all set to conqu er a world that is forbidden territory for the popular desktop OS Goliaths -- so hold your breath and wait for the fireworks to com e. They are sure to blow our mind. REFERENCES: [1] Personal communication, Ken Henry, GM Research. [2] http://www.wikipedi em, from Wikipedia, the free Encyclop edia [4] D. Cho and J.K. Hedrick, "Automotive Engine Modeling for Control", ASME Journal of Dynamic Systems, Measurement and Control, December 1989, Vol. 111, pp. 568-576. [5] Girard, A.R., Spry, S.C., Kretz, P.R. Dickey, S.R., Empey, D.M., Misener, J.A., Variaya, P.P. and Hedrick, J.K., "Vehicl e-to-Vehi cle Op en Experimental Platform Reference Manual. " [6]

Source: Ubiquity, Volume 6, Issue 28 (August 2 - August 9, 2005)