Abstract
In this paper, we propose a shared steering control strategy for semi-autonomous vehicles by taking into account the interaction between the driver and his/her vehicle. The copilot controller cooperates with the driver and help maintain the vehicle in the central position of a lane. Taking advantage of the small-gain theory, the design procedure does not rely on the perfect knowledge of the model and states for the driver. An adaptive dynamic programming method is introduced to develop a shared controller in real-time using online measurable input-output data from vehicle sensors. The efficacy of the proposed shared steering controller is demonstrated by computer-based simulations.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 155-160 |
| Number of pages | 6 |
| Journal | IFAC-PapersOnLine |
| Volume | 51 |
| Issue number | 13 |
| DOIs | |
| State | Published - Jan 1 2018 |
Keywords
- Cooperative driving
- adaptive dynamic programming (ADP)
- human in the loop
- steering control
ASJC Scopus subject areas
- Control and Systems Engineering
Fingerprint Dive into the research topics of 'Data-driven Shared Steering Control Design for Lane Keeping<sup>⁎</sup>'. Together they form a unique fingerprint.
Cite this
- APA
- Standard
- Harvard
- Vancouver
- Author
- BIBTEX
- RIS
Data-driven Shared Steering Control Design for Lane Keeping⁎ . / Huang, Mengzhe; Gao, Weinan; Wang, Yebin; Jiang, Zhong Ping.
In: IFAC-PapersOnLine, Vol. 51, No. 13, 01.01.2018, p. 155-160.Research output: Contribution to journal › Article › peer-review
}
TY - JOUR
T1 - Data-driven Shared Steering Control Design for Lane Keeping⁎
AU - Huang, Mengzhe
AU - Gao, Weinan
AU - Wang, Yebin
AU - Jiang, Zhong Ping
N1 - Funding Information: Data-driven Shared Steering Control Data-driven Shared Steering Control Data-Ddreisviegn SfohrarLeadneStKeeereipnigngC★★ontrol Design for Lane Keeping★ Design fo∗ r Lane Ke∗∗eping ∗∗∗ Mengzhe Huang ∗ Weinan Gao ∗∗ Yebin Wang ∗∗∗ Mengzhe Huang∗ Weinan Gao∗∗∗Yebin Wang∗∗∗ Mengzhe Huang Weinan Gao Yebin Wang MengzheHuanZgh∗oWng-PineinangGaoJian∗g∗∗Yebin Wang∗∗∗ Zhong-Ping Jiang∗ ∗ New York University, Brooklyn, NY 112fi1 USA ∗∗New York University, Brooklyn, NY 112fi1 USA New York University, Brooklyn, NY 112fi1 USA ∗ (e-mail: {m.huang, zjiang}@nyu.edu) ∗∗ New York University, Brooklyn, NY 112fi1 USA ∗∗Georgia Southern University, Statesboro, GA 3fi46fi USA Georgia S(oeu-mthaeriln: wUgnaiov@ergseitoyr,giSatsaotuetshbeorrno,.eGduA) 3fi46fi USA Georg∗∗ia∗ Southern University, Statesboro, GA 3fi46fi USA ∗∗∗Mitsubishi Electric Research Laboratories, Cambr∗∗∗idgeM,(ie-MtsmAubaiifsil2:h1iwgao@3E9leUctSrgAieco(Rreg-eimsaseaaouirlc:therhyeLban.ibnoewrdaaunt)ogr@ieise,ee.org) Cambr∗id∗∗ge, MA fi2139 USA (e-mail: yebinwang@ieee.org) Cambridge, MA fi2139 USA (e-mail: yebinwang@ieee.org) Cambridge, MA fi2139 USA (e-mail: yebinwang@ieee.org) Abstract: In this paper, we propose a shared steering control strategy for sefii-autonofious Abstract: In this paper, we propose a shared steering control strategy for sefii-autonofious Ψehicles Φy taking into account the interaction Φetweeeen the driΨer and his/her Ψehicle. The co-Abstract: In this paper, we propose a shared steering control strategy for sefii-autonofious pilot controller cooperates with the driΨer and help fiaintain the Ψehicle in the centtrraal position pilot controller cooperates with the driΨer and help fiaintain the Ψehicle in the central position of a lane. Taking adΨanttaage of the sfiall-gain theory, the design procedure does not rely on the pilot controller cooperates with the driΨer and help fiaintain the Ψehicle in the central position perfect knowledge of the fiodel and states for the driΨeerr.. Λn adaptiΨe dynafiic prografifiing fiethod is introduced to deΨelop a shared conttrrooller in real-tifie using online fieasuraΦle input-ofiuettphuotddisatiantrfroodfuiceΨdehtiocldeesΨeenlosporas.shTahreedefcfoicnatcryolloefr tinhereparl-otpifoiseedussinhgaroendlinsteeefriienagsucroanΦtlreoilnleprutis-output data frofi Ψehicle sensors. The efficacy of the proposed shared steering controller is output data frofi Ψehicle sensors. The efficacy of the proposed shared steering controller is douefiontputstdatrataedfrΦofiy cofipΨehiuclteersen-Φasedsors.siThfiuelatefficacyions. of the proposed shared steering controller is defionstrated Φy cofiputer-Φased sifiulations. d©ef2i0o1n8s, tIrFaAteCd (IΦnytecrnoaftiiponuatle rF-eΦdaesreatdiosnifoifuAlauttioomnast.ic Control) Hosting by Elsevier Ltd. All rights reserved. Keywords: CooperatiΨe driΨing, steering control, hufian in the loop, adaptiΨe dynafiic Keywords: CooperatiΨe driΨing, steering control, huuffiian in the loop, adaptiΨe dynafiic Keywords: CooperatiΨe driΨing, steering control, hufian in the loop, adaptiΨe dynafiic proKeygwraofirdfis:ingCoo(ΛpeDP).ratiΨe driΨing, steering control, hufian in the loop, adaptiΨe dynafiic prografifiing (ΛDP). prografifiing (ΛDP). 1. INTRODUCTION 1. INTRODUCTION 1. INTRODUCTION 1. INTRODUCTION Vehicle lane-keeping control intends to keep the Ψehicle in Vehicle lane-keeping control intends to keeeep the Ψehicle in VtVheehhiifcciliedlldaalneeo-kfeaeppciiunnrggΨicnognttorrroolsitnrtaeignnhddtss rtooakde.eVpattrhhioeeuΨsehhciioccnletriionnl Vehicle lane-keeping control intends to keep the Ψehicle in fiethods haΨe Φeen applied to design steering conttrroollers for fully autonofious Ψeehhicle (see, e.g., Cerone et al., 2009). fHfooorrwfuelΨlyera, uatonhouffiioauns ΨderhiΨieccrlle s(sstieeleel,, en.egge..d, sCetroonteaekteaaoll.Ψ,e22r0000t99h))e. for fully autonofious Ψehicle (see, e.g., Cerone et al., 2009). control, when the Ψeehhicle autofiation systefi encounters caHcoononnywttrrdeoloΨelef,,er,cwhwtihaoenennh.uSttfhhuieecahnΨΨeeadhhsriiuclciΨldeeedrauaenustttilltofiatorfainanestiiiteonoidsnonsysytofrstsotfefietaifikaeuenetnoocoucnΨoeourfnnittotheereurss any defection. Such a sudden transition frofi autonofious control, when the Ψehicle autofiation systefi encounters driΨing to fianuuaal driΨing is challenging and dangerous any defection. Such a sudden transition frofi autonofious for driΨer (Russell et al., 2016). In order to aΨoid the takee--driΨing to fianual driΨing is challenging and dangerous oΨer transition, seΨeral shared conttrrol strategies haΨe Φeen oΨer transition, seΨeral shared control strategies haΨe Φeen proposed to incorporate the hufian factor inttoo the design oΨer transition, seΨeral shared control strategies haΨe Φeen procedure, where the driΨer is constanttllyy steering the proposed to incorporate the hufian factor into the design Ψehicle along with the designed conttrrooller (see, e.g., Saleh Ψehicle along with the designed controller (see, e.g., Saleh et al., 2013; Gao et al., 2014). NotaΦly, there exist sofie Ψehicle along with the designed controller (see, e.g., Saleh ifiplefientation issues to achieΨe cooperation Φetweeeenn the et al., 2013; Gao et al., 2014). NotaΦly, there exist sofie driΨer and his/her Ψehicle. First, the driΨer’s fiodel can Φe ifiplefientation issues to achieΨe cooperation Φetween the different for different driΨeerrs. Second, the exact fiodel of a driΨer and his/her Ψehicle. First, the driΨer’s fiodel can Φe driΨer is cofiplex. Trraaditional fiodel-Φased fiethods do not driΨer is cofiplex. Traditional fiodel-Φased fiethods do not address Φoth the adaptiΨity and optifiality aspects. These arising lifiitations fiotiΨate our proposed data-driΨen non-afairroiissdiinnelgg-ΦlliiaffsiieiittdaattfiiiooenntsshfoidoottfiioΨΨraattteehoeuradpparrpooptioΨseedopddtaaitfaia--ddl rrdiiΨΨeesnignnono-f arising lifiitations fiotiΨate our proposed data-driΨen non-shared steering conttrrooller without accurate knowledge of fiodel-Φased fiethod for the adaptiΨe optifial design of the fiathefiatical fiodels for the driΨer and the Ψehicle. the fiathefiatical fiodels for the driΨer and the Ψehicle. ΛDP is a non-fiodel-Φased fiethod inspired Φy Φiological ΛDP is a non-fiodel-Φased fiethod inspired Φy Φiological learning and reinforcefient learning. It aifis to learn the loleepaatrrinnfiiinnagl caaonnnddtrrroeelinlfaowrrccefrfoifeintfilleearsnuirnagΦ..leIItinaapiiffuiitss-ottuootllpeeuaatrrnndtathhtaee optifial control law frofi fieasuraΦle input-output data owoppittthiiffoiiuaatll ctohnetreoxlalcatwknffrrooowffiiledfigeeasoufrasΦΦyllseeteifnipudtty--nooauufttippiuucstt (dsaetea, without the exact knowledge of systefi dynafiics (see, ★★wiThisthouwtortkhehasexbacteenksuppnowortedledgeinofpartsystbyefithedNyationalnafiicsficience(see, ★ This work has been supported in part by the National ficience FoTunhdisatwioonrkunhdaesrbGereanntsuEpCpCorfti-e1d50in10p44a,rtanbdy itnhepaNrtatbiyonaalgiffitcifernocme F★oundation under Grant ECCfi-1501044, and in part by a gift from FhoeThisunMdiattswiuoobnriksunderhnhasidEelrebcGteenrriacnRtsuppeEsCCfi-15010eaCortedrfcih-1L50ain1b0opart44,r4a,toandrnbiedys.inthenpaNrtationalbyagifficiencetfrom the Mitsubishi Electric Research Laboratories. Foundation under Grant ECCfi-1501044, and in part by a gift from 2405-8963 © 2018, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Proceedings, 2nd IFAC Conference on 155 PProeecre reedviinewgs u, n2dnedr IrFeAspCo nCsoibnilfietyre onfc Ien toenrnational Federation of Automa1t5ic5 Control. Proodceelelidnign,g Isd,e 2nntidfi cIFaAtiCon C aonndfe Creonnctreo ol nof Nonlinear 155 M1o0.d1e0l1li6n/gj.,i fIadceonl.t2i0fi1c8a.t0i7o.n27 a1nd Control of Nonlinear Myoosdcteeelmelidnsign,g Isd,e 2nntidfi cIFaAtiCon C aonndfe Creonnctreo ol nof Nonlinear 155 Systems SSystemsyuosadtdeealmlliansjga,r aId, eMnetixfiiccaot,i oJunn aen 2d0 C-2o2n, t2ro0l1 o8f Nonlinear Guadalajara, Mexico, June 20-22, 2018 Gyusatdeamlasjara, Mexico, June 20-22, 2018 Guadalajara, Mexico, June 20-22, 2018 e.g., Bian and Jiang, 2016; Jiang and Jiang, 2017; Gao e.g., Bian and Jiang, 2016; Jiang and Jiang, 2017; Gao e.g., JBiainagn, and Jiang, 2016; Jiang and Jiang, 2017; Gao and Jiang, 2017). Recently, ΛDP-Φased fiethods haΨe e.g., Bian and Jiang, 2016; Jiang and Jiang, 2017; Gao Φeen applied to design adaptiΨe optifial controllers for Φeen applied to design adaptiΨe optifial controllers for connected/autonofious Ψehicles (see, e.g., Gao et al., connected/autonofious Ψehicles (see, e.g., Gao et al., 2017;connectHuedan/aug ettonalofiou., 2017)s Ψ.ehicles (see, e.g., Gao et al., 2017; Huang et al., 2017). This paper presents a data-driΨen learning strategy to de-This paper presents a data-driΨen learning strategy to de-This paper presents a data-driΨen learning strategy to design a shared steering controller using ΛDP. First, Φased on This paper presents a data-driΨen learning strategy to de-the sfiall-gain theorefi applied to the interconnected sys-tehfeissfoifaldl-rgiΨaeinr athnedorΨeefhi calpe,ptliheed ptrootphoeseidntceorcoopnenraectitΨeed csoyns--tefis of driΨer and Ψehicle, the proposed cooperatiΨe con-the sfiall-gain theorefi applied to the interconnected sys-troller only uses the Ψehicle states as feedΦack without us-tefis of driΨer and Ψehicle, the proposed cooperatiΨe coning the internal states of the driΨer. Second, cofiΦined with inogΦuthste iandtaerpntaiΨlestadtyensaoffiitchepdrorigΨrearf.iSfeicinognd(,RcoΛfDiΦPin)e(dJwiainthg roΦust adaptiΨe dynafiic prografifiing (RΛDP) (Jiang ing the internal states of the driΨer. Second, cofiΦined with and Jiang, 2017), an iteratiΨe learning frafiework is deΨel-roΦust adaptiΨe dynafiic prografifiing (RΛDP) (Jiang oped to achieΨe lane keeping with fiinifiized lateral offset. TpheedftioaianchcioenΨterliaΦnuetikoenespainrge:w1i)ththfeiindiefsiiigznededlactoeroapleorafftsieΨte. The fiain contriΦutions are: 1) the designed cooperatiΨe oped to achieΨe lane keeping with fiinifiized lateral offset. Tohnetrfoilaleinr dcooenstrniΦouttdioenpsenadreo: n1)thteheundfeiseigansuedraΦcoleopinetreartniΨael controller does not depend on the unfieasuraΦle internal The fiain contriΦutions are: 1) the designed cooperatiΨe ctoanttersololefr tdhoeesdrniΨoetrd; e2p)enthdeosnhathreedunstfeieraisnugracΦolnetrinotlleernaisl states of the driΨer; 2) the shared steering controller is controller does not depend on the unfieasuraΦle internal learned frofi fieasuraΦle data of the Ψehicle, without the states of the driΨer; 2) the shared steering controller is exact knowledge of the fiathefiatical fiodels for the driΨer axnadctthkenoΨwehleicdlge.e of the fiathefiatical fiodels for the driΨer aexandcttheknoΨwehicledgle.eofthefiathefiaticalfiodelsforthedriΨer and the Ψehicle. T T T T Notations. Ψec(A) = [aT ,aT ,τ ττ, aT ]T , where a ∈ Notations. Ψec(A) = [a ,a ,τ ττ, a ] , where ai ∈ Notationn s. Ψec(A) = [aT,aT,τττ,n×amT]T, where a ∈ NRnotaarteionthse. Ψceoclu(Afi)ns=of[aA1,a∈2,τRτnτ,×amm.] |,·w|hreerpereasient∈s NRnotaaretionthes. Ψceoclufi(A)ns=of[aA1T,a∈2T,τRτnτ,×ammT.]T|,·wh|rerepereasient∈s R are the colufins of A1 ∈2 R m. | · | represients thne Euclidean norfi for Ψectors, orn×tmhe induced fiatrix R are the colufins of A ∈ R . | · | repremse×nmts nhoerfiEufocrlidfeiaantrinceosr.fiFofrora Ψseycftiofrise,troirc fthiaetriinxduσce∈d Rfima×trmix, norfi for fiatrices. For a syfifietric fiatrix σ ∈ Rm×m, the Euclidean norfi for Ψectors, or the induced fima×trmix nρoMrf(σi )foirs ftihaetrficieasx.ifFiourfiaesigyefnifΨiaelutreicoffiσat,riaxndσ Ψ∈ecsR(σ ) =, ρM(σ) is the fiaxifiufi eigenΨalue of σ, and Ψecs(mσ×)m= nρorfi(σ)forisfiattherfiaxices.ifFiourfiaeisygefifietnΨalureicoffiatσ,rianxdσΨT∈ecsR(σ) =, ρ[pM(,σ2p) is,·t·h·e,2fipaxif,ipufi,peig,e·n·Ψ·a,l2upe of σ,,apnd ]ΨTecs(σ) = ρ[pM11(,σ21p)m1i2(s,m·t+·he·1),2fiaxp1mif,piu22fi,pei23g,e·n··Ψa,l2upemof−1σρm,,panmmd ]ΨTTecs(σ) = [p11, 2221p12, · ·· , 2p1m,p22,p23, · ·· , 2pm−1ρm,pmm]T ∈[pn11R, 21pm12(,m·+1)·· ,.2pF1o2mr, pa2n2, pa2r3Φ, i·t·r·a,ry2pmcol2−1uρmfi,npmmΨec] tor 2ξT ∈ ∈Rn,RΨ21emcΨ((mξ+1)) =. [Fξo12r, ξ1aξn2, ·a·r·Φ,itξr1aξrny, ξc22o, ·lu· f· i,nξn−Ψe1cξtno,rξn2ξ]T ∈ ∈Rn1,RΨ2emcΨ((mξ+)1=). [Fξo12r,ξ1anξ2,·ar··Φ,ξitr1arξny,ξc2o,·lu··fin,ξn−Ψe1cξtnor,ξn2ξ]T ∈∈ RR2,,n(ΨΨnee+cc1ΨΨ()(ξξ)) == [[ξξ1,ξ,ξ1ξξ2,, ·· ···· ,ξ,ξ1ξξn,ξ,ξ2,, ·· ···· ,ξ,ξn−1ξξn,ξ,ξn]] ∈∈ Rn2n(n+1). 12 1 2 1 n 22 n−1 n n2T R1,n(Ψne+1)cΨ(.ξ) = [ξ ,ξ1ξ2,··· ,ξ1ξn,ξ ,··· ,ξn−1ξn,ξ ] ∈ R 21 n(n+1). 1 2 n R2n(n+1).
PY - 2018/1/1
Y1 - 2018/1/1
N2 - In this paper, we propose a shared steering control strategy for semi-autonomous vehicles by taking into account the interaction between the driver and his/her vehicle. The copilot controller cooperates with the driver and help maintain the vehicle in the central position of a lane. Taking advantage of the small-gain theory, the design procedure does not rely on the perfect knowledge of the model and states for the driver. An adaptive dynamic programming method is introduced to develop a shared controller in real-time using online measurable input-output data from vehicle sensors. The efficacy of the proposed shared steering controller is demonstrated by computer-based simulations.
AB - In this paper, we propose a shared steering control strategy for semi-autonomous vehicles by taking into account the interaction between the driver and his/her vehicle. The copilot controller cooperates with the driver and help maintain the vehicle in the central position of a lane. Taking advantage of the small-gain theory, the design procedure does not rely on the perfect knowledge of the model and states for the driver. An adaptive dynamic programming method is introduced to develop a shared controller in real-time using online measurable input-output data from vehicle sensors. The efficacy of the proposed shared steering controller is demonstrated by computer-based simulations.
KW - Cooperative driving
KW - adaptive dynamic programming (ADP)
KW - human in the loop
KW - steering control
UR - http://www.scopus.com/inward/record.url?scp=85052643382&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85052643382&partnerID=8YFLogxK
U2 - 10.1016/j.ifacol.2018.07.271
DO - 10.1016/j.ifacol.2018.07.271
M3 - Article
AN - SCOPUS:85052643382
VL - 51
SP - 155
EP - 160
JO - IFAC-PapersOnLine
JF - IFAC-PapersOnLine
SN - 2405-8963
IS - 13
ER -