A Novel Filter for Terrain Mapping With Laser Rangefinders
|
|
- Gwendoline Wheeler
- 5 years ago
- Views:
Transcription
1 Published i IEEE Trasacios o Roboics ad Auomaio, Vol., No. 5, Oc. 4, pp A Novel Filer for Terrai Mappig Wih Laser Ragefiders Cag Ye, Member, IEEE ad Joha Borese Member, IEEE Absrac This paper iroduces a ovel filer for errai mappig wih a -D laser ragefider. The filer, called Ceraiy Assised Spaial (CAS) filer, uses he physical cosrais o moio coiuiy ad spaial coiuiy o ideify corruped pixels ad missig daa i a elevaio map. The filer removes he corruped pixels, fills i he missig daa ad leaves he ucorruped pixels iac so as o preserve he deails of a errai map. Our exesive idoor ad oudoor mappig experimes show he CAS Filer s superior performace i erroeous daa reducio ad map deail preservaio over coveioal filers. Idex Terms laser ragefider, errai mappig, filerig, elevaio map, ceraiy map, mixed pixels. I. INTRODUCTION Auoomous avigaio of mobile robos o rugged errai requires he capabiliy o decide wheher a obsacle should be raversed or circumavigaed. The abiliy o make his decisio ad o acually execue i is called obsacle egoiaio (ON). A crucial issue ivolved i ON is errai mappig. Research effors o errai mappig have bee devoed o idoor eviromes [], oudoor, off-road errai [, 3, 4], as well as plaeary errai [5, 6, 7]. Mos of he exisig mehods employ sereovisio [, 3, 7], which is sesiive o eviromeal codiios (e.g., ambie illumiaio) ad has low rage resoluio ad accuracy. As a aleraive or suppleme, 3-D Laser Ragefiders (LRFs) have bee employed sice he early ieies [5, 6, 8]. However, 3-D LRFs are usually cosly, bulky, ad heavy. Therefore, hey are o suiable for small ad/or expedable robos. Furhermore, mos of hem are desiged for saioary use due o he slow frame rae. A more feasible soluio for lower-cos robos is a -D LRF. Researchers a CMU [] used a SICK -D LRF lookig upward o perform idoor 3-D mappig. The 3-D map is o suiable for ON sice o groud map is produced. Herikse ad Krokov [9] employed a Acuiy -D LRF o compleme sereovisio for obsacle ad hazard deecio. Due o he small look-ahead disace, he scaer was used for safey purposes oly (o sop he robo i case of a emergecy). More rece research a CMU [3] used a SICK LMS (wih a rage up o 3 m ad a field of view of ) o he Nomad robo. The LRF looks diagoally dowward ad forward. As he robo moves forward, he faig laser beam swep he groud ahead of he robo ad produced errai daa. Based o he daa, a so called goodess map, was creaed by comparig he curre sca daa wih he previous groud level (a leas square fi lie i he previous sca). The goodess map was he combied wih he map buil by sereovisio for pah plaig. I his work, he overall map was produced maily by sereovisio. This paper preses a ew errai mappig mehod usig a sigle sesor modaliy a -D LRF, ad a iovaive filerig mehod for map buildig. The paper is orgaized as follows: I Secio II, we iroduce our approach o errai mappig. I Secio III, we prese experimeal resuls ad discuss he compuaioal cos of he filer. I Secio IV, we compare our filer o oher relaed mixed pixel removal mehods ad discuss he real-ime implemeaio issues. II. MAP BUILDING AND FILTERING ALGORITHM A. Map buildig priciple As depiced i Fig., a SICK LMS is moued o our Gorilla vehicle such ha he LRF looks diagoally forward ad dowward a a pich agle of -. While he vehicle is i moio, he faig laser beam sweeps he errai ahead of he vehicle ad produces coiuous rage measuremes of he errai. For he experimes described i his paper, we se our LRF o a agular resoluio of ad a serial commuicaio speed of 5 Kbaud. Uder hese codiios, he LRF produces 8 rage readigs for is 8- degree field of view every 3.3 ms (see [] for deails). Our algorihm assumes ha real-ime, 6-degree-of-freedom pose iformaio is available from our so-called Propriocepive Pose Esimaio (PPE) sysem, which is described i deail i []. Wih he pose iformaio available, he rage daa ca be rasformed io world coordiaes ad regisered io a errai map. The errai map cosiss of a elevaio map ad a ceraiy map. Boh are -D grid-ype maps. Each cell i he elevaio map holds a value ha represes he heigh of he objec a ha Fig.. SICK LMS o he Gorilla vehicle cell, while each cell i he ceraiy map holds a value ha expresses our ceraiy i he accuracy of he correspodig cell i he elevaio map. I order o projec raw rage daa oo he -D elevaio map, we use he Euler rasformaio. The forward kiemaic rasformaio is give by: v v l d = Tl d () v v T l T where d = ( x, y, z ) ad d = ( xl, yl, zl ) are he rage measuremes i he world coordiae sysem ad i he LRF coordiae sysem, respecively. T l is he forward kiemaic rasformaio marix. For cociseess, we omi he derivaio of he rasformaio marix here. To sudy he mappig characerisics uder well-corolled codiios, we use he experimeal seup show i he lef of Fig.. The LRF is moued o a roary able, which produces roll ad pich roaio. The liear moio able x z o x b v θ q w o z b u m x l x s z l o r z s φ y l o3 y s laser β y b y Fig.. Experimeal seup ad he coordiae sysems for errai mappig sudy: he liear moio able moves he LRF a velociy v (max.: m/s) ad he roary able ils he LRF a a maximum rae of 37.5 /s i a rage of [-5, 5 ] for roll ad [-4, -5 ] for pich. produces liear moio for he roary able. This experimeal seup emulaes o some exe he moio of a mobile robo o rugged errai. The coordiae sysems for he experimeal seup are depiced i he righ of Fig.. They cosiss of he world coordiae sysem x y z, he coordiae sysems aached o he roary able x b y b z b ad x l y l z,, ad he LRF coordiae sysem x s y s z s. The coordiae values of o 3 i x l y l z l, o i x b y b z b, ad o i x y z are (,, p), (, m, q), ad (u, v, w), respecively. The forward kiemaic rasformaio is give by x y z lcφcβ + psφ + r + u = lcθsβ + lsθsφcβ qsθ + mcθ psθcφ + v lsθsβ lcθsφcβ + qcθ + msθ + pcθcφ + w where sx sads for si(x) ad cx sads for cos(x). β is he agle bewee he k h laser beam (which produces rage measureme l) ad he x s -axis ad is give by β = ( k ) π /8 (3) For every wo cosecuive measuremes a ime seps ad, he maximum chage of heigh value z for he measureme a (x, y ) is calculaed by z z z zmax = y aθ + δl + θ + φ (4) l θ φ ()
2 where y is he displaceme of he LRF alog he y axis, δ l is he maximum measureme error of he LRF (3 mm accordig o []), z / l, z / θ, ad z / φ are parial derivaives ha ca be derived from Eq.. I his paper, a grid cell size of 5 5 mm is used such ha each cell may be illumiaed oe ime eve if he vehicle moves a he maximum speed of m/s. A cell i he elevaio map is deoed h(, while a cell i he ceraiy map is deoed c(. Usig he above rasformaio he coordiae values x ad y are compued for each rage measureme a ime ad mapped o he grid idices i ad j, respecively. The, pixel c( i he ceraiy map is updaed as follows: c ( + a if z h ( z or (, ) + max c i j = = c ( (5) c ( oherwise where a is he icreme of he ceraiy value ad i ca be ay posiive ieger (we use 3 i his sudy). The pixel h( i he elevaio map is updaed by z if z > h ( h ( = (6) h ( oherwise I every 3.3-millisecod ierval he 8 rage measuremes acquired from he LRF are mapped io boh maps. B. Sources of mappig errors Ideally, a errai map buil accordig o he above-described algorihm should be sufficie for avigaig a mobile robo. However, some rage daa from he SICK LMS are erroeous ad will cause mappig errors. The pheomeo of mixed pixels, missig daa, ad arifacs/oise are he mai sources of erroeous measuremes: () Mixed pixels occur whe a laser beam his he very edge of a objec so ha par of he beam is refleced from ha edge of he objec ad par of he beam is refleced from he backgroud behid he objec. The resulig rage measureme lies somewhere bewee he disace o he objec ad he disace o he backgroud []. I researchig he mixed pixels pheomeo, we foud ha if he disace D bewee he edge of he foregroud objec ad he backgroud is close o he legh of he laser pulse, L ( L m for he SICK LRF ), a subsaial umber of mixed pixels are geeraed. However, if D > L, he umber of mixed pixels drops sigificaly (see Fig. 3). This is because he SICK LRF is desiged o accep oly reflecios semmig from he same pulse as valid daa. This smar desig is effecive i rejecig ambie oise. Fig. 3 shows ha mixed pixels exis eve if D is as big as.6 meer. Aoher characerisic of mixed pixels is ha a small chage i a edge s laeral posiio wih referece o he laser beam may iduce a large chage i he measured rage. () Missig daa occur whe he measured rage is ivalid []. For isace, o reur or oo weak a reured sigal may resul i missig daa; direc exposure o he suligh may lead o dazzlig ad cause ivalid readigs. For a complee lis of he SICK LRF s error codiios see []. (3) Eviromeal ierfereces, such as ambie ligh ad shock durig moio, may poeially creae oisy rage measuremes ad hece resul i oise i he elevaio map. The aalysis of LRF oise is reaed i [3]. Fig. 4 shows he effecs of hese errors. Such errors i he errai map deeriorae he ON performace of he vehicle. C. Coveioal image filerig algorihms A elevaio map ca be rasformed io a rage image, i which he heigh value of each pixel is represeed by iesiy. This way coveioal image processig filers [4, 5, 6] ca be applied o he elevaio map i a aemp o elimiae he above-discussed erroeous daa. However, he applicaio of coveioal image filerig echiques is o wihou drawbacks. The foremos reaso is ha mos coveioal image filerig mehods are ucodiioally applied o he eire rage image ad may hus blur he image. Covoluio-based frequecy filers, such as he Averagig Filer, Gaussia Filer [5]), or Wieer Filer [6], are implemeed Base o our commuicaio wih SICK i Germay, he LRF is a sigle sho measureme sysem. I seds ou a pulse wih a widh close o m ad expecs a refleced sigal wih he same pulse widh; received sigals of oher pulse widhs are rejeced. A a corer sho, he LRF may receive reflecio from boh he edge ad he backgroud if he disace bewee hem is wihi he pulse widh. This resuls i mixed pixels. by he covoluio of a ipu image wih a covoluio kerel i he spaial domai. The effec is ha oise is suppressed bu he objecs i he image are Fig. 3. Characerisics of mixed pixels: 4 rage measuremes were ake a each poi. A measureme is deermied as a mixed pixel if he differece bewee he measured rage ad he foregroud/backgroud is bigger ha 6 mm. Large umbers of mixed pixels are geeraed whe he disace D bewee he edge of he foregroud objec ad he backgroud is close o he legh of a pulse of he laser, L. However, if D > L, he he umber of mixed pixels drops sigificaly. (a) Mixed pixels ca creae phaom objecs behid he edges of objecs. Missig daa creaes empy pixels i he upper surfaces of objecs. (b) Eviromeal ierfereces ca cause arifacs/oise. I he 6-meer measurig mode of he SICK LRF, we observed ha here are a lo of high spikes (labeled arifacs ) i he ceer of his errai map, which was buil o a fla sidewalk ear our lab. Close examiaio revealed ha mos of hese arifacs resuled from very small rage measuremes, which may be removed by hreshold. We suspec ha udeeced overflow is he cause. The oher high spikes are he resul of radom oise i rage daa, which produces rage errors a several meers ahead of he LRF. Fig. 4. Map misrepreseaio due o rage errors blurred a he same ime. Filers based o Aisoropic Diffusio [7] diffuse more i smooh areas ad less aroud large iesiy rasiios, hus small variaios i image iesiy (i.e., hose ha are caused by oise) are smoohed ad edges are preserved. However, mixed pixels may creae large phaom objecs, which look like edges raher ha like oise, alog he edges of objecs. I may be difficul o smooh hem based o he gradie of iesiy wihou diffusig he edges. Media Filers have he advaage of o iroducig ay ew pixel io he images, hereby preservig deails. Media Filers have bee used o remove mixed pixels i laser rage images [4]. The sadard Media Filer ad is variaios (Weighed Media Filer [8] ad Ceer Weighed Media Filer [9]) ofe exhibi blurrig whe a large filer widow is used, or
3 3 isufficie oise reducio for a small filer widow. Adapive Media Filers [, ] maiai a beer balace bewee deail preservaio ad oise reducio, ad hece achieve beer performace. However, all of hese media-ype filers affec all he pixels i a image icludig ucorruped oes. A umber of media ype filers i he scieific lieraure [, 3] may poeially disiguish corruped pixels from ucorruped oes ad apply filerig oly o he corruped pixels. These filers uilize local ipu image characerisics o ideify corruped pixels. However, a limiaio of his approach is he poeial misclassificaio of hi edges as corruped pixels ad he subseque removal of hese pixels. For isace, a hi pole i a rage image produces a lie of high iesiy pixels, which may be misierpreed by he filer as impulse oise ad may herefore be removed. Based o his discussio, we summarily coclude ha i is isufficie o use oly spaial iformaio for reliable ideificaio ad removal of corruped pixels. max D. The Ceraiy Assised Spaial (CAS) Filer I his paper, we propose a ovel filer for elevaio map buildig. The filer is called Ceraiy Assised Spaial (CAS) filer. This filer uilizes o oly he spaial iformaio coaied i he ufilered elevaio map, bu also he ceraiy iformaio coaied i he ceraiy map, as explaied i Secio II. A. Due o he coiuiy of moio, every wo cosecuive measuremes o he edge of a real objec saisfy z h ( zmax. Accordig o Eq. 5, edge pixels are assiged coiuous ceraiy icremes ad resul i large ceraiy values i he ceraiy map. Therefore, edge pixels i he elevaio map ca be ideified easily. They are preserved ad o filerig is applied. Mixed pixels may occur somewhere bewee he edges of objecs ad he backgroud. Their locaios are deermied by he perceage of laser spos hiig he backgroud []. Whe he LRF is i moio, as i ecessarily is for mappig, his perceage chages all he ime. Cosequely, mixed pixels keep chagig heir x- ad y-coordiaes all he ime. This meas ha he codiio z h ( z is o saisfied. Hece, mixed pixels usually have small ceraiy value ad are spaially isolaed i boh he elevaio map ad he ceraiy map. They ca be ideified ad removed easily. Arifacs ad radom oise have he same characerisics ad ca also be removed by he same priciple. Fially, missig daa creae empy pixels i a elevaio map ad heir ceraiy values are zero. We foud ha applyig a Weighed Media Filer o hose pixels wih zero ceraiy values ca supply (ierpolae) he missig daa. The filerig process works as follows: Le H={h( i M, j N} ad C={c( i M, j N} deoe he elevaio map ad he ceraiy map, respecively, where M is he legh ad N is he widh of he map. A filer wih a widow size of (k+) (k+) (i his paper we use k=) is defied symmerically surroudig he curre pixel h(i, H. The curre pixel h( is updaed by he oupu of he filer, which is give by if c( A ad gs( = ad gc( = y( = ywm ( else if gs( = ad c( = (7) h( oherwise ad A = a (i( s / v ) + ) (8) where i( ) meas roud o he eares ieger, s is he grid cell size, v is he liear velociy of he LRF (he maximum is m/s i his paper) ad =3.3 ms is a ime sep. I his equaio, i( s / v ) is he esimae o he umber of imes a grid is illumiaed. I order o filer he mixed pixels a he viciiy of he edges, a cosa value of is added. g s ( ad g c ( are he spaial coiuiy idex of H ad C for he pixels i he filer widow, respecively. They are expressed by if σ ( > B h g s ( = (9) oherwise ad i+ k, j < > = + k if c(m,) G or σc( E g c ( () m= i-k, = j k oherwise where σ ad σ h c are he variace of he ormalized local maps H L ad C L ormalized by he maximum elevaio ad ceraiy values, respecively. We use A for G ad.4 for E. To keep he compuaioal cos low we i k, j use + + k H L < 7 o approximae he codiio σ ( > B i Eq. 9. y ( h wm is he i- j- j- j j+ j+ m= i k, = j k i- i oupu of a Weighed Media (WM) filer i+ ceered a pixels (. The ipu pixels o he WM filer are obaied by applyig a filer i+ skeleo [8] (as depiced i Fig. 5) o he Fig. 5. Filer skeleo for he elevaio map. For isace, h( has oe proposed WM filer copy, while h(i-, j-) has wo copies i he ipu pixels o he WM filer. The use of his skeleo is based o he followig heurisic: Sice we suspec h( o be a missig daum, he eares eighborig pixels h(i±, j±) are also likely o be missig daa, while hose pixels farher from h( have a greaer likelihood o o be missig daa. Noe, however, ha his weigh assigme for he WM filer poeially removes more missig daa o he upper surface of a objec. Whe he CAS Filer is applied o he raw elevaio map, mixed pixels, arifacs, ad radom oise are ideified by examiig he moio coiuiy (i.e., c( cosraied by Eq. 5) ad he spaial coiuiy i he elevaio map ad he ceraiy map ( g s ( ad g c ( i, ). Missig daa is disiguished by ispecig he curre pixel s ceraiy value c( ad he spaial coiuiy g s ( i he elevaio map. Mixed pixels, arifacs, ad radom oise are herefore removed ad he missig daa are filled by he WM filer. III. EXPERIMENTAL RESULTS I his secio, we prese exesive experimeal resuls. These resuls ca be grouped io wo caegories: Bechmark ess performed uder highly corolled codiios, ad mobiliy ess i which he LRF was moued o a mobile plaform. I all of he experimes, we used a. GHz AMD Ahlo processor-based PC ruig RT-Liux for he real-ime daa collecio. A. Bechmark Tess Based o he experimeal seup i Fig., we desiged he followig four bechmark ess: ) The LRF raslaes oly (deoed T ); ) he LRF raslaes ad roaes wih roll oly (deoed TR ); 3) he LRF raslaes ad roaes wih pich oly (deoed TP ); ad 4) he LRF raslaes ad roaes wih roll ad pich (deoed TRP ). For he experimes described here, we buil 6 obsacle courses as lised i Table I (e.g., obsacle course #6 comprises parallelepiped &, he cylider, pole -4 ad pole 6). The performace of our CAS Filer is compared wih ha of he Wieer Filer, he Averagig Filer, he Media Filer, ad he Ceer Weighed Media (CWM) Filer. Fig. 6 shows he 4 h obsacle course, while Fig. 7 depics he raw ad filered elevaio maps based o bechmark es T wih differe filers. I ca be observed ha he CAS Filer preserves more origial pixels, removes mos of he mixed pixels, ad fill almos all of he missig daa. The Wieer Filer reais almos all missig daa, ad some mixed pixels, bu i oly smoohes some oher mixed pixels. The Media Filer ad he CWM Filer over-filer he poles while he Averagig Filer disor he map sigificaly. Examiaio o all oher ess reveals ha he CAS filer preserved all poles (as small as.5 mm i diameer) while successfully removig he mixed pixels. Fig. 6. Oe of he 6 idoor obsacle courses.
4 4 Table I: Feaures of he 6 obsacle courses Obsacle Parallelepiped Cylider Poles layou 3 4 X X -5, 7, 9 X -7, 9 3-7, 9 4 X X X,, 4, 6 5 X X -4, 6 6 X X X -4, 6 7 X X X -7 8 X X X -9 9 X X -9 X X X -5, 7, 9 X X X -5, 7, 9 X X -5, 7, 9 3 X X -7, 9 4 X X X 5 X X X 6 X X Dimesios of parallelepiped objecs (all uis i mm): #: , #: , #3: , #4: Dimesios of poles: #-#9 have diameers of 4,, 9, 5,, 9, 6, 4.5 ad.5, respecively ad a legh of 94. Dimesio of cylider: diameer=4, heigh= 459 For each of he 6 obsacle courses, we performed four differe ess, amely, he T, TR, TP, ad he TRP es, as described i he begiig of his secio. The resuls are summarized i Fig. 8. I is evide from hese resuls ha our proposed CAS Filer cosisely ouperformed he four coveioal filers. The relaively poor performace of he Media Filer ad he CWM Filer is due o heir ihere over-filerig. O he oher had, whe roll ad pich moio was added o he raslaory moio, he CAS Filer performed oly margially beer ha he Wieer Filer (ad he Averagig Filer i some isaces). This is due o he followig observaios: () Roll ad/or pich moio poeially icreased mappig errors because hey caused vibraios of he roary able ad chages i occlusios. Accordig o Eq., hese addiioal errors move he CAS Filer s PI close o ha of he Wieer ad Averagig filer. () Roll ad pich moio may icrease he amou of empy pixels o he objecs upper surfaces. Whe he umber of empy pixels i he filer widow is above a cerai value, he CAS Filer removes he curre pixel, while he Wieer Filer or he Averagig Filer replace he curre pixel wih a spaial covoluio value. However, he deerioraio of he CAS Filer s PI does o meas ha he Wieer Filer ad Averagig Filer are more suiable for ON. This is because he Wieer Filer does o remove arifacs/oise i elevaio maps (see Secio III. B), while he Averagig Filer eds o lower he heigh of edges i he elevaio maps a highly udesirable rai. I order o compare he filer performace i a quaiaive way, we measured by had he exac locaio of each obsacle i each obsacle course ad covered hese measuremes io a groud ruh elevaio map. We he evaluae he performace of filer by compuig a overall idex of performace (PI) accordig o N M j= i= PI = N M ( h ( h ( ) ( h ) r ( h ( j= i= f () where h r (, h ( ad h f ( are he elevaio values of pixel ( i he raw elevaio map, he groud ruh map, ad he filered map, respecively. We arbirarily seleced M=8 ad N= (which has full coverage of all obsacle courses) o evaluae he experimes here. Accordig o Eq., a smaller PI meas beer filer performace. Noe: We foud ha he iclusio of σ c ( >. 4 i he bechmark ess did o improve he filer s performace sigificaly. Therefore, we dropped his erm o save compuaio i ON applicaios. B. Experimeal resuls o a mobile robo We carried ou a umber of idoor ad oudoor map buildig experimes wih he LRF o our Gorilla vehicle (see Fig. ). However, hese experimes were limied o sraigh-lie moio o fla, horizoal groud. More diverse errai would require real-ime pose daa from our PPE sysem, which we have o ye iegraed wih he ON applicaio described here. Fig. 9 shows he obsacle courses ad he vehicle i our idoor ad oudoor mappig experimes. From he maps depiced i Fig., i is appare ha he CAS Filer ouperforms he Wieer Filer because i almos compleely removes mixed pixels (a) Origial (raw) elevaio map (b) Elevaio map afer applyig he CAS Filer (c) Elevaio map afer applyig he Wieer Filer (d) Elevaio map afer applyig he Averagig Filer (e) Elevaio map afer applyig he Media Filer (f) Elevaio map afer applyig he CWM filer Fig. 7. Origial elevaio map ad he maps afer applyig he five esed filers: he liear moio able moved he LRF alog he direcio v
5 5 3.5 PI.5.5 Wieer Average Media.5.5 PI.5 W ieer Average Media PI (a) Traslaio oly (c) Traslaio wih roll Fig. 8. Compariso of filer performace W ieer Average Media Fig. 9. s for mappig: lef: idoor, righ: oudoor ad fills mos of he missig daa o he op surfaces of objecs (see Fig. (b))..5.5 PI (b) Traslaio wih pich (d) Traslaio wih roll & pich W ieer Average Media I coras, he Wieer Filer leaves some oiceable mixed pixels ad mos of he missig daa i he map, as show i Fig. (c). I he oudoor mappig experime (see Fig. ), he CAS Filer compleely removed all arifacs ad radom oise, while he Wieer Filer did o remove ay. C. Preservaio of deails ad compuaioal cos The preservaio of deails ad compuaioal cos of he CAS Filer depeds o each case ad we are uable o provide a geeral aalyical aalysis. However, we ca give a prelimiary esimae based o he bechmark ess. Table II shows he perceage of pixels ideified by he CAS Filer as mixed pixels ad missig daa i bechmark es T. For he wors case (he 3rd obsacle course) i he 6 obsacle courses, 74.8% of pixels were lef iac. This meas ha he CAS Filer preserved a sigifica amou of deails, (a) Origial (raw) elevaio map (b) Elevaio map afer applyig he CAS Filer (c) Elevaio map afer applyig he Wieer Filer Fig.. Idoor elevaio maps: he vehicle moved alog direcio v (a) Origial (raw) elevaio map (b) Elevaio map afer applyig he CAS Filer (c) Elevaio map afer applyig he Wieer Filer Fig.. Oudoor elevaio maps: he vehicle moved alog direcio v
6 6 TABLE II PERCENTAGE OF PIXELS IDENTIFIED AS CORRUPTED IN BENCHMARK TEST T s Mixed pixels Missig daa No filerig Noe: The oal umber of pixels i he maps is, of which oly pixels are processed by he CAS Filer. which is also he mai moive for he developme of he CAS Filer. For a esimae o compuaioal cos, we compared he CAS Filer wih he sadard Media Filer. From Equaios (7)-(), we ca deermie ha i order o process he pixels i a row (K=96), he CAS Filer requires: P+K- addiios, K- subracios, 3K muliplicaios ad K divisios. The algorihm furher icurs complexiy.76 KQ lgq i he WM filer ad K(P-) i compariso (o fid he maximum). P=5 ad Q=45 are he umber of pixels i he filer widows ad he umber of pixels i he WM filer skeleo (Fig. 5), respecively. The Media Filer has a complexiy of KPlg P. For ieger operaio i he PC used i he experime, we foud: () he four fudameal operaios of arihmeic ad he sorig operaio have roughly he same compuaioal cos,.7 s; ad () a compariso operaio has a cos of.85 s. Cosiderig he worse case (he 3rd obsacle course), he esimaed average compuig ime of he CAS Filer is. ms, while ha of he Media Filer is 38.7 ms. This meas ha he CAS Filer is less expesive i compuaio ha he Media Filer. Alhough i requires addiioal compuaio i buildig he ceraiy map (Eq. 4 & 5), his is o sigifica. Thus we believe ha he CAS Filer is compuaioally more efficie ha he Media Filer. A examiaio of he oher bechmark ess revealed similar resuls, which we omi here because of space limiaios. IV. DISCUSSION A. Compariso o oher mixed pixels removal mehods We are aware ha here is a aleraive o filerig mixed pixels. Adams [4] proposed a efficie mixed pixels removal mehod for phase-shif LRFs. The algorihm is based o he deecio of discoiuiy of he received sigal s ampliude. However, due o he differece of measurig priciples (imeof-fligh i he case of he SICK LRF), his mehod mus be modified o sui he SICK LRF. Such modificaio requires kowledge of he ieral physics ad he hardware desig of he LRF, ad ca hus be implemeed effecively oly by he maufacurer, o he user. Specifically, Adam s discoiuiy deecio algorihm requires several samples of he received sigal (for compuig he secod derivaive of he square of he sigal ampliude wih respec o he area illumiaed by he laser o he edge) durig he ime he laser beam raverses he edge. However, a user level, we ca oly ge a sigle measureme o he edge. This meas ha i is impracical o impleme his algorihm. Fially, our mehod is based o he characerisics of he erroeous daa ad he moio coiuiy cosrai ad i is hus device idepede. Therefore, i is more geeral ad suiable for he ed users of LRFs. I coras o Adams algorihm, our approach removes o oly mixed pixels bu also uexpeced oudoor arifacs ad oise. B. Real-ime implemeaio issues The CAS Filer ca be used for mappig where off-lie processig is sufficie. For real-ime obsacle egoiaio, he CAS Filer ca be implemeed as follows: wihi he compuer s world model, a so-called acive widow of size 4 pixels (rows colums) ravels wih he vehicle. The LRF is locaed a he ceer of he firs row. I each LRF samplig period, oly he pixels (i boh elevaio & ceraiy map) from he (-m) h row o he h row ahead of he vehicle are updaed. The he pixels i row -m is filered ad regisered io aoher elevaio map, which is used for obsacle egoiaio. m is chose i such a way ha he vehicle ca acquire sufficie obsacle/errai iformaio i elevaio. The grid cell size of he elevaio map for obsacle egoiaio may be bigger ha 5 5 mm subjec o applicaio. Our errai mappig mehod requires real-ime vehicle pose iformaio, which will be obaied by our PPE sysem. Therefore, i is o ecessary o perform simulaeous localizaio ad mappig i our mehod. The PPE sysem was proposed i [] ad he researchers a our lab have already buil his sysem. We will iegrae his sysem io our mappig mehod i he immediae fuure. V. CONCLUSIONS I his paper, we preseed a ovel filer for errai mappig wih a -D LRF. The errai map cosiss of a elevaio map ad a ceraiy map. The value of each pixel i he ceraiy map is deermied as a fucio of a socalled moio coiuiy cosrai. Typical erroeous daa, such as mixed pixels, arifacs, ad radom oise do saisfy he moio coiuiy cosrai ad hus resul i a low ceraiy value. Examiaio of he appropriae pixels i boh maps allows our CAS Filer o ideify mixed ad missig pixels. This disicio, i ur, allows he CAS Filer o apply differe reames o hese differe codiios. Ucorruped pixels are lef iac o preserve he map deails. Overall, our proposed CAS Filer produces cosisely beer resuls ha coveioal filerig algorihms because i makes use of he addiioal iformaio available i he ceraiy map. ACKNOWLEDGMENT This work was fuded by he U.S. Deparme of Eergy uder Award No. DE-FG4-86NE3796 ad uder a gra from The Uiversiy of Michiga s Auomoive Research Ceer (ARC), fuded by TACOM. The auhors would like o hak Mr. Weber, SICK Germay for his kid help ad isighful advise o he LRF. The auhors would also like o hak James Berry for his work o he desig of he roary able. REFERENCES S. Thru, W. Burgard ad D. Fox, A real-ime algorihm for mobile robo mappig wih applicaio o muli-robo ad 3D mappig, Proc. IEEE I. Cof. Roboics ad Auomaio,, pp S. Begé-Brezez, e al, Ucerai map makig i aural eviromes, Proc. IEEE I. Cof. Roboics ad Auomaio, 996, pp D. S. Aposolopoulos, e al, Techology ad field demosraio of roboic search for Aarcic meeories, I. J. Roboics Research, vol. 9, o., pp. 5-3,. 4 K. Fregee, R. Madhava ad L. E. Parker, Icremeal muli-age roboic mappig of oudoor errai, Proc. IEEE I. Cof. Roboics ad Auomaio,, pp I. S. Kweo ad Takeo Kaade, High-resoluio errai map from muliple sesor daa, IEEE Tras. Paer Aal. Machie Iell., vol. 4, o., pp. 78-9, E. Krokov ad R. Hoffma, Terrai mappig for a walkig plaeary rover, IEEE Tras. Robo. Auoma., vol., o. 6, pp , R. Simmos, e al, Experiece wih rover avigaio for luar-like errais, Proc. IEEE/RSJ I. Cof. Iellige Robos ad Sysem, 995, pp C. M. Shoemaker ad J. A. Borse The demo III UGV program: a esbed for auoomous avigaio research, Proc. IEEE ISIS/CIRA/ISAS Joi Coferece, 998, pp L. Herikse ad E. Krokov, Naural Terrai Hazard Deecio wih a Laser Ragefider, Proc. IEEE I. Cof. Roboics ad Auomaio, 997, pp C. Ye ad J. Borese Characerizaio of a -D laser scaer for mobile robo obsacle egoiaio, Proc. IEEE I. Cof. Roboics ad Auomaio,, pp L. Ojeda ad J. Borese "FLEXav: Fuzzy Logic Exper Rule-based Posiio Esimaio for Mobile Robos o Rugged Terra" Proc. IEEE I. Cof. Roboics ad Auomaio,, pp SICK Ic., LMS/LMI Telegram Lisig, versio 5., page D. Niza, e al, The measureme ad use of regisered reflecace ad rage daa i scee aalysis, Proceedigs of he IEEE, vol. 65, o., M. Heber ad E. Krokov, 3D measuremes from imagig laser radars: how good are hey? Image ad Visio Compuig, vol., o. 3, pp. 7-78,99. 5 C. F. Olso, Adapive-scale filerig ad feaurig deecio usig rage daa, IEEE Tras. Paer Aal. Machie Iell., vol., o. 9,. 6 A. K. Ja Fudameals of Digial Image Processig, Preice Hall, P. Peroa ad J. Malik, Scale-space ad edge deecio usig aisoropic diffusio, IEEE Tras. Paer Aal. Machie Iell., vol., o. 7, D. R. K. Browrigg, The weighed Media Filer, Commuicaio of he ACM, vol. 7, o. 8, pp , 984.
7 9 S. Ko ad S. Lee, Ceer weighed Media Filers ad heir applicaios o image ehaceme, IEEE Tras. Circuis ad Sysems, vol. 38, o. 9, pp , 99. H. Li ad A. N. Willso, Jr., Media Filers wih adapive legh, IEEE Tras. Circuis ad Sysems, vol. 35, o. 6, 988. T. Che ad H. R. Wu, Applicaio of pariio-based media ype filers for suppressig oise i images, IEEE Tras. Image Processig, vol., o. 6, pp ,. H. Eg ad K. Ma, Noise adapive sof-swichig Media Filer, IEEE Tras. Image Processig, vol., o., pp. 4-5,. 3 T. Che, K. Ma ad L. Che, Tri-sae Media Filer for image deoisig, IEEE Tras. Image Processig, vol. 8, o., pp , M. D. Adams ad P. J. Prober, The ierpreaio of phase ad iesiy daa from AMCW ligh deecio sesors for reliable ragig, I. J. Robo. Res., vol. 5, No. 5, pp ,
A new terrain mapping method for mobile robots obstacle negotiation
A ew errai mappig mehod for mobile robos obsacle egoiaio Cag Ye ad Joha Boresei Advaced Techologies Lab The Uiversiy of Michiga 0 Beal Ave, A Arbor, MI, USA 4809-0 ABSTRACT This paper iroduces a ew errai
More informationIntroduction to Orthogonal Frequency Division Multiplexing (OFDM) Technique
Wireless Iformaio Trasmissio Sysem Lab. Iroducio o Orhogoal Frequecy Divisio Muliplexig (OFDM Techique Isiue of Commuicaios Egieerig Naioal Su Ya-se Uiversiy OFDM Overview OFDM Sysem Model Orhogoaliy Oulie
More informationIntroduction to Orthogonal Frequency Division Multiplexing (OFDM)
Wireless Iformaio Trasmissio Sysem Lab. Iroducio o Orhogoal Frequecy Divisio Muliplexig OFDM Isiue of Commuicaios Egieerig aioal Su Ya-se Uiversiy OFDM Overview OFDM Sysem Model Orhogoaliy Oulie Muli-carrier
More informationIntroduction to OFDM Systems
Wireless Iformaio Trasmissio Sysem Lab. Iroducio o OFDM Sysems Isiue of Commuicaios Egieerig Naioal Su Ya-se Uiversiy Oulie OFDM Overview OFDM Sysem Model Orhogoaliy Muli-carrier Equivale Implemeaio by
More informationIn Color We Live C IP E I. In Color We Live. t n. In Color We Live
I r We Live Eviro me I r We Live IP E I 0 12 I 2 T E vir ome 0 12 I 2 T IPEI Evirome IPEI 012 I 2 T I r We Live Iesiy Idepede RGB-o-XYZ ur amera alibraio Bria FUNT, ad Pouya BSTNI School of ompuig Sciece,
More informationWalsh Function Based Reconstruction Method of. Command Waveform in PWM Inverter
Walsh Fucio Based Recosrucio Mehod of Commad Waveform i PWM Iverer Kri Choeisai ad Seiji Kodo Nagaoka Uiversiy of echology 63- Kamiomioka, Nagaoka 94-288, Japa Fa: +8-28-4-9, Phoe: +8-28-4-9 E-Mail: kri@s.agaokau.ac.jp
More informationHandout 10: Pulse-Code Modulation
ENGG 31-B: Priciples of Commuicaio Sysems Hadou 1: Pulse-Code Modulaio 17 18 Firs Term Isrucor: Wig-Ki Ma November 1, 17 Suggesed Readig: Chaper 7 of Simo Hayki ad Michael Moher, Commuicaio Sysems (5h
More informationComparison of Color Features for Image Retrieval
Compariso of Color Feaures for Image Rerieval S.R. Kodiuwakku 1 Deparme of Saisics & Compuer Sciece, Uiversiy of Peradeiya salukak@pd.ac.lk S.Selvarajah Deparme of Physical Sciece, Vavuiya Campus, Uiversiy
More informationExperiment 2: Waveform Analysis
Specrum aalyzer Experime : Waveform Aalysis 8/3/6 his experime is o familiarize you wih he aalysis of sigals ad eworks usig he HP 358A Specrum Aalyzer. he aalysis of sigals is achieved i he ope loop mode
More informationLimb-Vector Paralleling: A General Approach to Translate Postures from Human to Humanoid Robots
Limb-Vecor Parallelig: A Geeral Approach o Traslae Posures from Huma o Humaoid Robos Zhuo-Li Bao, Shih-Li Che, Chi-Chag Tug, Rayi Yau Tara, Hadziq Fabroyir ad Wei-Chug Teg Absrac evelopig iuiive ierface
More informationA New Power System Oscillation Type Identification Method Based on Empirical Mode Decomposition and Hilbert Transform
Proceedigs of he World Cogress o Egieerig ad Compuer Sciece 4 Vol I WCECS 4, -4 Ocober, 4, Sa Fracisco, USA A New Power Sysem Oscillaio Type Ideificaio Mehod Based o Empirical Mode Decomposiio ad Hilber
More informationSelf-Interference Cancellation in Full-Duplex Radio Transceivers with Oscillator Phase Noise
Self-Ierferece Cacellaio i Full-Duplex Radio Trasceivers wih Oscillaor Phase Noise Ville Syrjälä ad Koji Yamamoo Graduae School of Iformaics Kyoo Uiversiy Yoshida-homachi Sakyo-ku Kyoo-shi 606-8501 Kyoo
More informationAn optimization algorithm for the calculation of the electrostatic discharge current equations parameters
WSEAS TRANSACTIONS o CIRCUITS ad SYSTEMS Vasiliki Via George P. Fois Lambros Ekoomou A opimizaio algorihm for he calculaio of he elecrosaic discharge curre equaios parameers VASILIKI VITA GEORGE P. FOTIS
More informationAn Optimization of Gaussian UWB Pulses
0 h Ieraioal Coferece o DEVELOPMENT AND APPLICATION SYSTEMS, Suceava, Romaia, May 7-9, 00 A Opimizaio of Gaussia UWB Pulses Adria POPA "Gheorghe Asachi" Techical Uiversiy of Iaşi Bd.Carol No., RO0.506
More informationField-Effect Transistors (FETs) 3.4 The p-channel MOSFET & COMS
Field-Effec Trasisors (FETs 3.4 The p-chael MOSFET & COMS A p-chael ehaceme-ype MOSFET is fabricaed o a -ype subsrae wih p+ regios for he drai ad source, ad has holes as charge carriers. The deice operaes
More informationITERATIVE RECEIVER FOR SPACE-TIME CODED FLAT FADING CHANNELS. Noura Sellami yz, Mohamed Siala y, David Declercq z, Inbar Fijalkow z
ITERATIVE RECEIVER FOR SPACE-TIME CODED FLAT FADING CHANNELS Noura Sellami yz, Mohamed Siala y, David Declercq z, Ibar Fijalkow z y Frace Telecom R&D, 38-40 rue du Gééral Leclerc 92794 Issy les Moulieaux,
More informationMULTI USER DETECTION FOR CDMA- OFDM/OQAM SYSTEM COMBINED WITH SPACE TIME CODING
Ieraioal Joural of Wireless & Mobile Neworks (IJWMN) Vol. 4, No. 4, Augus 212 MULTI USER DETECTION FOR CDMA- OFDM/OQAM SYSTEM COMBINED WITH SPACE TIME CODING Radhia GHARSALLAH 1 ad Ridha BOUALLEGUE 2 1
More informationDetection of Pulsed Radar in a Time Division Duplexed System
Deecio of Pulsed Radar i a Time Divisio Duplexed Sysem Brad W. Zarikoff Hamilo Isiue Naioal Uiversiy of Irelad Mayooh Email: Brad.Zarikoff@uim.ie David Weldo Wireless R&D, Vecima Neworks, Ic. Vicoria,
More informationExpanding the Twin Pulse Swing Free Profile
8 IEEE Ieraioal Coferece o Roboics ad Auomaio Pasadea, CA, USA, May 19-3, 8 Expadig he Twi Swig Free Profile David Bowlig, Gregory Sarr, Joh Wood ad Ro Lumia Deparme of Mechaical Egieerig The Uiversiy
More informationBézier curves with shape parameter *
Wag e al. / J Zhejiag Uiv SCI 5 6A(6):497-5 497 Joural of Zhejiag Uiversiy SCIENCE ISSN 9-395 hp://www.zju.edu.c/jzus E-mail: jzus@zju.edu.c Bézier curves wih shape parameer * WANG We-ao ( 王文涛 ) WANG Guo-zhao
More informationA New Adaptive Medical Image Coding with Orthogonal Polynomials
Ieraioal Joural of Compuer Applicaios (975 8887) Volume 46 No., May A New Adapive Medical Image Codig wih Orhogoal Polyomials Krishamoorhy R Deparme of CSE Aa Uiversiy of Techology - Tiruchirappalli Tiruchirappalli,
More informationGeo-Encryption Protocol For Mobile Networks
Geo-Ecrypio Proocol For Mobile Neworks Ala Al-Fuqaha, Omar Al-Ibrahim, Ammar Rayes * Deparme of Compuer Sciece Weser Michiga Uiversiy Kalamazoo, MI {alfuqaha, oalibrah } @cs.wmich.edu Absrac We propose
More informationEE123 Digital Signal Processing
Aoucemes EE3 Digial Sigal Processig Lecure Miderm: Friday ex week Ope everyhig... bu chea shee recommeded isead Who ca o say ill pm? Opioal homework ex week Will give you miderm ad pracice quesios How
More informationFast and Accurate Behavioral Simulation of Fractional-N Frequency Synthesizers and other PLL/DLL Circuits
Fas ad Accurae Behavioral Simulaio of Fracioal-N Frequecy Syhesizers ad oher PLL/LL Circuis Michael H. Perro Microsysems Techology Laboraory, MIT perro@mi.edu, hp://www-ml.mi.edu/ perro ABSTRACT Techiques
More informationON THE QAM PARALLEL TURBO-TCM SCHEMES USING RECURSIVE CONVOLUTIONAL GF(2 N ) ENCODERS
8h Europea Sigal Processig Coferece (EUSIPCO-00) Aalborg, Demark, Augus 3-7, 00 O THE QAM PARAE TURBO-TCM SCHEMES USIG RECURSIVE COVOUTIOA GF( ) ECODERS Adria Flori Pău, Cǎli Vlǎdeau, Io Marghescu, Safwa
More informationJune : 2016 (CBCS) Negative terminal for forward bias. Positive terminal for forward bias. Cathode ( n-type)
s Year : Commo o all raches Jue : 6 (CCS) Noe : Max. marks : 6 (i) (ii) (iii) (iv) emp ay five quesios ou of seve quesios. ll quesios carry equal marks. Draw he ea diagram, wheever ecessary. ssume daa,
More informationTransmit Power Efficiency of Multi-Hop Hybrid Selection/MRC Diversity for a DS-CDMA Virtual Cellular Network
Trasmi ower Efficiecy of Muli-Hop Hybri Selecio/ Diversiy for a DS-CDMA Virual Cellular ework Imae DAOU, Eisuke KUDOH a Fumiyuki ADACHI Dep. of Elecrical a Commuicaio Egieerig, Grauae School of Egieerig,
More informationMoment-Based Bound on Peak-to-Average Power Ratio and Reduction with Unitary Matrix
Mome-Based Boud o Peak-o-Average Power Raio ad Reducio wih Uiary Marix Hirofumi Tsuda Deparme of Applied Mahemaics ad Physics Graduae School of Iformaics Kyoo Uiversiy Kyoo Japa suda.hirofumi.38u@s.kyoo-u.ac.jp
More informationResearch on Reliability for Wind Turbine Transmission System Based on Random Theory
Advaced ciece ad Techology Leers Vol8 (CT 5) pp3-9 hp://ddoiorg/457/asl58 Research o Reliabiliy or Wid Turbie Trasmissio ysem Based o Radom Theory Xi Gua Huadog Wag 3 Zhili u Da Zhao chool o Mechaical
More informationPerformance Analysis of an Ultra-Tightly Integrated GPS and Reduced IMU System
Performace Aalysis of a Ulra-Tighly Iegraed GPS ad Reduced IMU Sysem M.G. Peovello, D. Su, G. Lachapelle, M.E. Cao Posiio, Locaio ad Navigaio (PLAN) Group Deparme of Geomaics Egieerig Schulich School of
More informationDigital Communications (by B. Sklar)
Digial Commuicaios by B. Sklar Sug Ho Cho Voice 2 222 39 Cellular 11 412 5178 E mail: drago@hayag.ac.kr Coes 1 Sigals ad Specra Formaig ad Basebad Trasmissio Badpass Modulaio ad Demodulaio Chael Codig
More informationPower-Efficient Radio Resource Allocation for Low- Medium-Altitude Aerial Platform Based TD-LTE Networks
ower-efficie Radio Resource Allocaio for Low- Medium-Aliude Aerial laform Based TD-LTE eworks Liqiag Zhao, 2, Chi Zhag, Haili Zhag ad Xiaohui Li. Sae Key Laboraory of Iegraed Service eworks, Xidia Uiversiy,
More informationR. W. Erickson. Department of Electrical, Computer, and Energy Engineering University of Colorado, Boulder
R. W. Erickso Deparme of Elecrical, Compuer, ad Eergy Egieerig Uiversiy of Colorado, Boulder 4.2.2. The Power MOSFET Gae Source Gae leghs approachig oe micro p - p Cosiss of may small ehacememode parallelcoeced
More informationA Channel Estimation Method for MIMO-OFDM Mobile WiMax Systems
A Chael Eimaio Mehod for MIMO-OFDM Mobile WiMax Syem Fabie Delere ad Yichuag Su School of Egieerig ad echology Uiveriy of Herfordhire, Hafield, Her, AL0 9AB, UK Abrac I hi paper, chael eimaio for Space-ime
More informationMultiuser Precoding for UWB Sensor Networks
Muliuser Precodig for UWB Sesor Neworks Heirich Luecke ad Armi Wiebe Commuicaio Techology Laboraory, ETH Zurich, 89 Zurich, Swizerlad {lueckeh, wiebe}@ari.ee.ehz.ch Absrac We cosider he dowlik of a UlraWidebad
More informationMIMO-OFDM with Pilot-Aided Channel Estimation for WiMax Systems
Fir Ieraioal Workhop o he Performace Ehaceme i MIMO-OFDM Syem MIMO-OFDM wih Pilo-Aided Chael Eimaio for WiMax Syem Fabie Delere ad Yichuag Su School of Egieerig ad echology Uiveriy of Herfordhire, Hafield,
More informationImproving the energy efficiency for the WBSN bottleneck zone based on random linear network coding
Improvig he eergy efficiecy for he WBSN boleeck zoe based o radom liear ework codig Al Shahee, H ad Takruri Rizk, H hp://dx.doi.org/10.1049/ie wss.2017.0056 Tile Auhors Type URL Improvig he eergy efficiecy
More informationPhase-Shifting Control of Double Pulse in Harmonic Elimination Wei Peng1, a*, Junhong Zhang1, Jianxin gao1, b, Guangyi Li1, c
Inernaional Symposium on Mechanical Engineering and Maerial Science (ISMEMS 016 Phase-Shifing Conrol of Double Pulse in Harmonic Eliminaion Wei Peng1, a*, Junhong Zhang1, Jianxin gao1, b, Guangyi i1, c
More informationPROJECT #2 GENERIC ROBOT SIMULATOR
Uiversity of Missouri-Columbia Departmet of Electrical ad Computer Egieerig ECE 7330 Itroductio to Mechatroics ad Robotic Visio Fall, 2010 PROJECT #2 GENERIC ROBOT SIMULATOR Luis Alberto Rivera Estrada
More informationPower Conversion and Signal Transmission Integration Method Based on Dual Modulation of DC-DC Converters
IEEE TRANSATIONS ON INDUSTRIAL ELETRONIS 1 Power oversio ad Sigal Trasmissio Iegraio Mehod Based o Dual Modulaio of D-D overers Jiade Wu, Member, IEEE, Ji Du, Sude Member, IEEE, Zhegyu Li, Seior Member,
More information2600 Capitol Avenue Suite 200 Sacramento, CA phone fax
26 Capiol Avenue Suie 2 Sacrameno, CA 9816 916.64.4 phone 916.64.41 fax www.esassoc.com memorandum dae Sepember 2, 216 o from subjec Richard Rich, Ciy of Sacrameno; Jeffrey Dorso, Pioneer Law Group Brian
More information(This lesson plan assumes the students are using an air-powered rocket as described in the Materials section.)
The Mah Projecs Journal Page 1 PROJECT MISSION o MArs inroducion Many sae mah sandards and mos curricula involving quadraic equaions require sudens o solve "falling objec" or "projecile" problems, which
More informationIEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 32, NO. 5, MAY
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 32, NO. 5, MAY 2014 1 Updae-Efficiecy ad Local Repairabiliy Limis for Capaciy Approachig Codes Arya Mazumdar Member, IEEE, Veka Chadar, ad Gregory
More informationTECHNIQUE FOR TRACKING AND VISUALIZATION OF MOTION IN SEQUENCE OF IMAGES OF THE SOLAR CORONA
Proc. VI Serbia-ulgaria Asroomical Coferece elgrade 7-11 Ma 2008 Eds. M. S. Dimirijevi M. Tsveov L.. Popovi V. Golev Publ. Asr. Soc. "Rudjer ošovi" No. 9 2009 207-216 TECHNIQUE FOR TRACKING AND VISUALIZATION
More informationAPPLICATION NOTE UNDERSTANDING EFFECTIVE BITS
APPLICATION NOTE AN95091 INTRODUCTION UNDERSTANDING EFFECTIVE BITS Toy Girard, Sigatec, Desig ad Applicatios Egieer Oe criteria ofte used to evaluate a Aalog to Digital Coverter (ADC) or data acquisitio
More informationMeasurement of Equivalent Input Distortion AN 20
Measuremet of Equivalet Iput Distortio AN 2 Applicatio Note to the R&D SYSTEM Traditioal measuremets of harmoic distortio performed o loudspeakers reveal ot oly the symptoms of the oliearities but also
More informationG.711.1: A WIDEBAND EXTENSION TO ITU-T G.711
16h Europea Sigal Processig Coferece (EUSIPCO 008), Lausae, Swizerlad, Augus 59, 008, copyrigh by EURASIP G.711.1: A WIDEBAND ETENSION TO ITUT G.711 Yusue Hiwasai 1, Shigeai Sasai 1, Hioshi Ohmuro 1, Taeshi
More informationSpring Localization I. Roland Siegwart, Margarita Chli, Martin Rufli. ASL Autonomous Systems Lab. Autonomous Mobile Robots
Spring 2017 Localizaion I Localizaion I 10.04.2017 1 2 ASL Auonomous Sysems Lab knowledge, daa base mission commands Localizaion Map Building environmen model local map posiion global map Cogniion Pah
More informationCHAPTER 5 A NEAR-LOSSLESS RUN-LENGTH CODER
95 CHAPTER 5 A NEAR-LOSSLESS RUN-LENGTH CODER 5.1 GENERAL Ru-legth codig is a lossless image compressio techique, which produces modest compressio ratios. Oe way of icreasig the compressio ratio of a ru-legth
More informationNegative frequency communication
Negaive frequency communicaion Fanping DU Email: dufanping@homail.com Qing Huo Liu arxiv:2.43v5 [cs.it] 26 Sep 2 Deparmen of Elecrical and Compuer Engineering Duke Universiy Email: Qing.Liu@duke.edu Absrac
More informationLab 3 Acceleration. What You Need To Know: Physics 211 Lab
b Lab 3 Acceleraion Wha You Need To Know: The Physics In he previous lab you learned ha he velociy of an objec can be deermined by finding he slope of he objec s posiion vs. ime graph. x v ave. = v ave.
More informationIN SOUTHERN AFRICA MICHAELGOCHFELD
Codor 83:243251 0 The Cooper Omihologd Sociey 1981 NEST SITE SELECTION IN SOUTHERN AFRICA BY KELP GULLS JOANNA AND BURGER MICHAELGOCHFELD ABSTRACT.We sudied six coloies of Kelp Gulls (Larus domiicaus)
More informationLecture September 6, 2011
cs294-p29 Seminar on Algorihmic Game heory Sepember 6, 2011 Lecure Sepember 6, 2011 Lecurer: Chrisos H. Papadimiriou Scribes: Aloni Cohen and James Andrews 1 Game Represenaion 1.1 abular Form and he Problem
More informationMemorandum on Impulse Winding Tester
Memorandum on Impulse Winding Teser. Esimaion of Inducance by Impulse Response When he volage response is observed afer connecing an elecric charge sored up in he capaciy C o he coil L (including he inside
More informationResearch Article Improving the Performance of Modular Production in the Apparel Assembly: A Mathematical Programming Approach
Mahemaical Problems i Egieerig, Aricle ID 472781, 7 pages hp://dx.doi.org/1.1155/214/472781 Research Aricle Improvig he Performace of Modular Producio i he Apparel Assembly: A Mahemaical Programmig Approach
More informationKeywords:. Consumer Acceptance, Purchasing Power, Root Mean Square Error, Theil U Statistics
Ieraioal Joural of Ecoomics, Busiess ad Maageme Research Vol., No. 05; 07 ISSN: 456-7760 APPLICATION OF THE LOTKA-VOLTERRA MODEL IN EXPLORING THE SALES VOLUMES INTERACTIONS OF SMARTPHONES BETWEEN ASIA
More informationEECE 301 Signals & Systems Prof. Mark Fowler
EECE 3 Signals & Sysems Prof. Mark Fowler Noe Se #8 C-T Sysems: Frequency-Domain Analysis of Sysems Reading Assignmen: Secion 5.2 of Kamen and Heck /2 Course Flow Diagram The arrows here show concepual
More informationModeling and Prediction of the Wireless Vector Channel Encountered by Smart Antenna Systems
Modeling and Predicion of he Wireless Vecor Channel Encounered by Smar Anenna Sysems Kapil R. Dandekar, Albero Arredondo, Hao Ling and Guanghan Xu A Kalman-filer based, vecor auoregressive (VAR) model
More informationMEASUREMENTS OF VARYING VOLTAGES
MEASUREMENTS OF ARYING OLTAGES Measuremens of varying volages are commonly done wih an oscilloscope. The oscilloscope displays a plo (graph) of volage versus imes. This is done by deflecing a sream of
More information<Diode Modules> RM200CY-24S HIGH POWER SWITCHING USE INSULATED TYPE
dual swich (Cahode Common) Forward curren...... 2 0 0 A Repeiive peak reverse volage V RRM... 1 2 0 0 V Maximum juncion emperaure T jmax... 1 5 0 C Fla base Type Copper base plae RoHS Direcive
More informationPointwise Image Operations
Poinwise Image Operaions Binary Image Analysis Jana Kosecka hp://cs.gmu.edu/~kosecka/cs482.hml - Lookup able mach image inensiy o he displayed brighness values Manipulaion of he lookup able differen Visual
More informationImplementation of Variable Duty Cycle Ring Oscillator for Unique Identification
Imlemeaio of Variable Duy Cycle Rig Oscillaor for Uique Ideificaio N.Sivasakari (), B.Balambiga () ad P.Idumahi (3) () Assisa Professor, Dearme of Elecroics ad Commuicaio Egieerig Meco Schlek college of
More informationECE3204 Microelectronics II Bitar / McNeill. ECE 3204 / Term D-2017 Problem Set 7
EE3204 Microelecronics II Biar / McNeill Due: Monday, May 1, 2017 EE 3204 / Term D-2017 Problem Se 7 All ex problems from Sedra and Smih, Microelecronic ircuis, 7h ediion. NOTES: Be sure your NAME and
More informationFuzzy Inference Model for Learning from Experiences and Its Application to Robot Navigation
Fuzzy Inference Model for Learning from Experiences and Is Applicaion o Robo Navigaion Manabu Gouko, Yoshihiro Sugaya and Hiroomo Aso Deparmen of Elecrical and Communicaion Engineering, Graduae School
More informationRevision: June 11, E Main Suite D Pullman, WA (509) Voice and Fax
2.5.3: Sinusoidal Signals and Complex Exponenials Revision: June 11, 2010 215 E Main Suie D Pullman, W 99163 (509) 334 6306 Voice and Fax Overview Sinusoidal signals and complex exponenials are exremely
More informationNoise Reduction/Mode Isolation with Adaptive Down Conversion (ADC)
Page 1 Noise Reducion/Mode Isolaion wih Adapive Down Conversion (ADC) Abel B. Diaz, Thomas W. Tunnell NSTec Los Alamos Operaions Presened o PDV Workshop 8-16-2007 Page 2 Summary Adapive down conversion
More informationP. Bruschi: Project guidelines PSM Project guidelines.
Projec guidelines. 1. Rules for he execuion of he projecs Projecs are opional. Their aim is o improve he sudens knowledge of he basic full-cusom design flow. The final score of he exam is no affeced by
More informationAn off-line multiprocessor real-time scheduling algorithm to reduce static energy consumption
An off-line muliprocessor real-ime scheduling algorihm o reduce saic energy consumpion Firs Workshop on Highly-Reliable Power-Efficien Embedded Designs Shenzhen, China Vincen Legou, Mahieu Jan, Lauren
More informationLecture 11. Digital Transmission Fundamentals
CS4/MSc Compuer Neworking Lecure 11 Digial Transmission Fundamenals Compuer Neworking, Copyrigh Universiy of Edinburgh 2005 Digial Transmission Fundamenals Neworks consruced ou of Links or ransmission
More informationECMA st Edition / June Near Field Communication Wired Interface (NFC-WI)
ECMA-373 1 s Ediion / June 2006 Near Field Communicaion Wired Inerface (NFC-WI) Sandard ECMA-373 1 s Ediion / June 2006 Near Field Communicaion Wired Inerface (NFC-WI) Ecma Inernaional Rue du Rhône 114
More informationImproved Performance for Color to Gray and Back For Orthogonal transforms using Normalization
Ieraioal Joral of Compaioal Egieerig Research Vol 3 Isse 5 Improved Performace for Color o Gray ad Bac For Orhogoal rasforms sig ormalizaio Dr. H. B. Kere Dr. Sdeep D. Thepade Raesh. Charvedi 3 Sr. Prof.
More informationIs Now Part of To learn more about ON Semiconductor, please visit our website at
s Now Par of To lear ore abou ON Seicoducor, please visi our websie a www.osei.co ON Seicoducor ad he ON Seicoducor logo are radearks of Seicoducor poes dusries, C dba ON Seicoducor or is subsidiaries
More informationJeremiah: The Face of Computer Vision
Jeremiah: The Face of Compuer Visio Richard Bowde CVSSP, School of ECM Guildford Surrey +44 (0)1483 689838 r.bowde@surrey.ac.uk ABSTRACT This paper preses a humaoid compuer ierface (Jeremiah) ha is capable
More informationLogarithms APPENDIX IV. 265 Appendix
APPENDIX IV Logarithms Sometimes, a umerical expressio may ivolve multiplicatio, divisio or ratioal powers of large umbers. For such calculatios, logarithms are very useful. They help us i makig difficult
More informationVariation Aware Cross-Talk Aggressor Alignment by Mixed Integer Linear Programming
ariaion Aware Cross-alk Aggressor Alignmen by Mixed Ineger Linear Programming ladimir Zoloov IBM. J. Wason Research Cener, Yorkown Heighs, NY zoloov@us.ibm.com Peer Feldmann D. E. Shaw Research, New York,
More informationElectrical connection
Reference scanner Dimensioned drawing en 02-2014/06 50117040-01 200 500mm Disance on background/reference 10-30 V DC We reserve he righ o make changes DS_HRTR46Bref_en_50117040_01.fm Robus objec deecion
More informationCommunications II Lecture 7: Performance of digital modulation
Communicaions II Lecure 7: Performance of digial modulaion Professor Kin K. Leung EEE and Compuing Deparmens Imperial College London Copyrigh reserved Ouline Digial modulaion and demodulaion Error probabiliy
More informationTechnology Trends & Issues in High-Speed Digital Systems
Deailed comparison of dynamic range beween a vecor nework analyzer and sampling oscilloscope based ime domain reflecomeer by normalizing measuremen ime Sho Okuyama Technology Trends & Issues in High-Speed
More informationMotion-blurred star image acquisition and restoration method based on the separable kernel Honglin Yuana, Fan Lib and Tao Yuc
5h Inernaional Conference on Advanced Maerials and Compuer Science (ICAMCS 206) Moion-blurred sar image acquisiion and resoraion mehod based on he separable kernel Honglin Yuana, Fan Lib and Tao Yuc Beihang
More informationImage Denoising using Fast Bilateral Filter
Iteratioal Joural of Curret Egieerig ad Techology E-ISS 77 4106, P-ISS 347 5161 015IPRESSCO, All Rights Reserved Available at http://ipressco.com/category/ijcet Research Article Riku A Joseph * ad Maju
More information5 Spatial Relations on Lines
5 Spaial Relaions on Lines There are number of useful problems ha can be solved wih he basic consrucion echniques developed hus far. We now look a cerain problems, which involve spaial relaionships beween
More informationDisribued by: www.jameco.com 1-800-831-4242 The conen and copyrighs of he aached maerial are he propery of is owner. 16K-Bi CMOS PARALLEL E 2 PROM FEATURES Fas Read Access Times: 200 ns Low Power CMOS
More informationModulation exercises. Chapter 3
Chaper 3 Modulaion exercises Each problem is annoaed wih he leer E, T, C which sands for exercise, requires some hough, requires some concepualizaion. Problems labeled E are usually mechanical, hose labeled
More informationMean Waiting and Turnaround Time for Multiple Web Transaction
Mea Waiig ad uraroud ie for Muliple Web rasacio Yog-Ji Lee Dep. of Copuer Sciece Woosog Uiversiy aejo, 300-78, Korea yjlee@woosog.ac.kr M. Aiquzzaa School of Copuer Sciece Uiversiy of Oklahoa Nora, OK
More informationEE368/CS232 Digital Image Processing Winter Homework #1 Released: Monday, January 8 Due: Wednesday, January 17, 1:30pm
EE368/CS232 Digial Image Processing Winer 207-208 Lecure Review and Quizzes (Due: Wednesday, January 7, :30pm) Please review wha you have learned in class and hen complee he online quiz quesions for he
More informationEE 40 Final Project Basic Circuit
EE 0 Spring 2006 Final Projec EE 0 Final Projec Basic Circui Par I: General insrucion 1. The final projec will coun 0% of he lab grading, since i s going o ake lab sessions. All oher individual labs will
More informationDirect Analysis of Wave Digital Network of Microstrip Structure with Step Discontinuities
Direc Analysis of Wave Digial Nework of Microsrip Srucure wih Sep Disconinuiies BILJANA P. SOŠIĆ Faculy of Elecronic Engineering Universiy of Niš Aleksandra Medvedeva 4, Niš SERBIA MIODRAG V. GMIROVIĆ
More informationEE201 Circuit Theory I Fall
EE1 Circui Theory I 17 Fall 1. Basic Conceps Chaper 1 of Nilsson - 3 Hrs. Inroducion, Curren and Volage, Power and Energy. Basic Laws Chaper &3 of Nilsson - 6 Hrs. Volage and Curren Sources, Ohm s Law,
More information16.5 ADDITIONAL EXAMPLES
16.5 ADDITIONAL EXAMPLES For reiew purposes, more examples of boh piecewise linear and incremenal analysis are gien in he following subsecions. No new maerial is presened, so readers who do no need addiional
More informationTehrani N Journal of Scientific and Engineering Research, 2018, 5(7):1-7
Available olie www.jsaer.com, 2018, 5(7):1-7 Research Article ISSN: 2394-2630 CODEN(USA): JSERBR 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38
More informationThe University of Melbourne Department of Mathematics and Statistics School Mathematics Competition, 2013 JUNIOR DIVISION Time allowed: Two hours
The Universiy of Melbourne Deparmen of Mahemaics and Saisics School Mahemaics Compeiion, 203 JUNIOR DIVISION Time allowed: Two hours These quesions are designed o es your abiliy o analyse a problem and
More informationEE 330 Lecture 24. Amplification with Transistor Circuits Small Signal Modelling
EE 330 Lecure 24 Amplificaion wih Transisor Circuis Small Signal Modelling Review from las ime Area Comparison beween BJT and MOSFET BJT Area = 3600 l 2 n-channel MOSFET Area = 168 l 2 Area Raio = 21:1
More informationAn Adaptive Image Denoising Method based on Thresholding
A Adaptive Image Deoisig Method based o Thresholdig HARI OM AND MANTOSH BISWAS Departmet of Computer Sciece & Egieerig Idia School of Mies, Dhabad Jharkad-86004 INDIA {hariom4idia, matoshb}@gmail.com Abstract
More informationSolution of ECE 342 Test 2 S12
Soluion of ECE 342 Tes 2 S2. All quesions regarding superheerodyne receivers refer o his diagram. x c () Anenna B T < B RF < 2 f B = B T Oher Signals f c Mixer f Baseband x RFi RF () x RFo () () () x i
More informationSimulation Analysis of DC-DC Circuit Based on Simulink in Intelligent Vehicle Terminal
Open Access Library Journal 218, Volume 5, e4682 ISSN Online: 2333-9721 ISSN Prin: 2333-975 Simulai Analysis of DC-DC Circui Based Simulink in Inelligen Vehicle erminal Weiran Li, Guoping Yang College
More informationEXPERIMENT #9 FIBER OPTIC COMMUNICATIONS LINK
EXPERIMENT #9 FIBER OPTIC COMMUNICATIONS LINK INTRODUCTION: Much of daa communicaions is concerned wih sending digial informaion hrough sysems ha normally only pass analog signals. A elephone line is such
More informationRadar emitter recognition method based on AdaBoost and decision tree Tang Xiaojing1, a, Chen Weigao1 and Zhu Weigang1 1
Advaces i Egieerig Research, volume 8 d Iteratioal Coferece o Automatio, Mechaical Cotrol ad Computatioal Egieerig (AMCCE 7) Radar emitter recogitio method based o AdaBoost ad decisio tree Tag Xiaojig,
More information10. The Series Resistor and Inductor Circuit
Elecronicsab.nb 1. he Series esisor and Inducor Circui Inroducion he las laboraory involved a resisor, and capacior, C in series wih a baery swich on or off. I was simpler, as a pracical maer, o replace
More informationMEDIUM VOLTAGE CABLES TO IEC60502, BS6622, VDE0276
IEC2, BS22, VDE2 Sigle Core Cables o VDE 2 The sigle core cables are desiged for disribuio of elecrical power wih Applicaio: omial volage Uo/U ragig from./.kv o 9/KV ad frequecy Hz. They are suiable for
More informationEXPERIMENT #4 AM MODULATOR AND POWER AMPLIFIER
EXPERIMENT #4 AM MODULATOR AND POWER AMPLIFIER INTRODUCTION: Being able o ransmi a radio frequency carrier across space is of no use unless we can place informaion or inelligence upon i. This las ransmier
More information