Lateral Model Predictive Control for Autonomous Heavy-Duty Vehicles
Total Page：16
File Type：pdf, Size：1020Kb
kth royal institute of technology Licentiate Thesis in Electrical Engineering Lateral Model Predictive Control for Autonomous Heavy-Duty Vehicles Sensor, Actuator, and Reference Uncertainties GONÇALO COLLARES PEREIRA Stockholm, Sweden 2020 Lateral Model Predictive Control for Autonomous Heavy-Duty Vehicles Sensor, Actuator, and Reference Uncertainties GONÇALO COLLARES PEREIRA Academic Dissertation which, with due permission of the KTH Royal Institute of Technology, is submitted for public defence for the Degree of Licentiate of Engineering on Tuesday the 15th September 2020 at 10:00 a.m. in Harry Nyquist at KTH Royal Institute of Technology, Malvinas väg 10, Stockholm. Licentiate Thesis in Electrical Engineering KTH Royal Institute of Technology Stockholm, Sweden 2020 © Gonçalo Collares Pereira ISBN: 978-91-7873-580-8 TRITA-EECS-AVL-2020:38 Printed by: Universitetsservice US-AB, Sweden 2020 Abstract Autonomous vehicle technology is shaping the future of road transportation. This technology promises safer, greener, and more efficient means of transportation for everyone. Autonomous vehicles are expected to have their first big impact in closed environments, such as mining areas, ports, and construction sites, where heavy-duty vehicles (HDVs) operate. Although research for autonomous systems has boomed in recent years, there are still many challenges associated with them. This thesis addresses lateral motion control for autonomous HDVs using model predictive control (MPC). First, the autonomous vehicle architecture and, in particular, the control module architecture are introduced. The control module receives the current vehicle states and a trajectory to follow, and requests a velocity and a steering-wheel angle to the vehicle actuators. Moreover, the control module needs to handle system delays, maintain certain error bounds, respect actuation constraints, and provide a safe and comfortable ride. Second, a linear robust model predictive controller for disturbed discrete-time nonlinear systems is presented. The optimization problem includes the initial nom- inal state of the system, which allows to guarantee robust exponential stability of the disturbance invariant set for the discrete-time nonlinear system. The controller effectiveness is demonstrated through simulations of an autonomous vehicle lateral control application. Finally, the controller limitations and possible improvements are discussed with the help of a more constrained autonomous vehicle example. Third, a path following reference aware MPC (RA-MPC) for autonomous vehi- cles is presented. The controller makes use of the linear time-varying MPC frame- work, and considers control input rates and accelerations to account for limitations on the vehicle steering dynamics and to provide a safe and comfortable ride. More- over, the controller includes a method to systematically handle references generated by motion planners which can consider different algorithms and vehicle models from the controller. The controller is verified through simulations and through experi- ments with a Scania construction truck. The experiments show an average lateral error to path of around 7 cm, not exceeding 27 cm on dry roads. Finally, the nonlinear curvature response of the vehicle is studied and the MPC prediction model is modified to account for it. The standard kinematic bicycle model does not describe accurately the lateral motion of the vehicle. Therefore, by extending the model with a nonlinear function that maps the curvature response of the vehicle to a given request, a better prediction of the vehicle’s movement is achieved. The modified model is used together with the RA-MPC and verified through simulations and experiments with a Scania construction truck, where the improvements of the more accurate model are verified. The experiments show an average lateral error to path of around 5 cm, not exceeding 20 cm on wet roads. Sammanfattning Autonoma fordon f¨orv¨antas f˚aen stor inverkan p˚aframtidens transporter av gods och personer. En teknologi som lovar s¨akrare,gr¨onareoch effektivare transporter till alla. Den typ av verksamhet som autonoma fordon f¨orstf¨orv¨antas f˚aett st¨orregenomslag inom ¨artransporter i avskilda omr˚aden,s˚asom gruvomr˚aden,hamnar och byggplatser. Aven¨ om forskning kopplat till autonoma system har exploderat under den senaste ˚aren kvarst˚arfortfarande ett flertal fr˚agest¨allningar. Denna avhandling fokuserar p˚alateral r¨orelsestyrning av tunga autonoma fordon med modellprediktiva regulatorer (MPC). Avhandlingen best˚arav fyra huvuddelar. I f¨orstdelen introduceras det autonoma fordonets systemarkitektur, med fokus p˚aregulatormodulen. Regulatormodulen gener- erar hastighet och rattvinkel referenser till fordonets hastighetaktuator och rattvinkelak- tuator baserat p˚afordonets nuvarande tillst˚andsamt den givna referensbanan som for- donet skall f¨olja.Regulatormodulen beh¨over dessutom hantera f¨ordr¨ojningari systemet, s¨akerst¨allaatt systemet inte ¨overskrider givna felmarginaler, hantera aktuator och system- begr¨ansningar,och sist men inte minst framf¨orafordonet p˚aett s¨akert och komfortabelt s¨att. I andra delen presenteras en robust modellprediktiv regulator f¨or ett tidsdiskret olinj¨art system med st¨orningar.I optimeringsproblemet inkluderas systemets nominella initialtill- st˚and,detta m¨ojligg¨orgaranterad robust exponentiell stabilitet f¨ordet tidsdiskreta olinj¨ara systemets st¨orningsinvarianta tillst˚andsm¨angd.Regulatorns prestanda visas genom simu- leringar av ett autonomt fordon d¨arregulatorn kontrollerar fordonets laterala r¨orelse. Begr¨ansningaroch potentiella f¨orb¨attringarav regulatorn diskuteras utifr˚anexempel med ¨okadebegr¨ansningar. I tredje delen presenteras en referens medveten modellprediktiv regulator (RA-MPC), en regulator utvecklad f¨oratt styra ett autonomt fordon l¨angsen given referensbana. Regulator baseras p˚aen linj¨arttidsvarierande MPC och begr¨ansningari fordonets styrdy- namik hanteras genom att ber¨aknadessa baserat p˚ain insignalernas, referensbana, v¨arden och derivator. Genom att beakta begr¨ansningarnap˚adetta s¨att m¨ojligg¨orsen komfortabel och s¨aker k¨orning.En systematisk metod f¨oratt hantera referensbanor som genererats av r¨orelseplanerarebaseras p˚aalgoritmer och modeller som skiljer sig fr˚ande som anv¨ands i regulatorn presenteras ocks˚a.Den metoden ¨ar¨aven implementerad i regulatorn. Regu- latorn har utv¨arderatsmed s˚av¨alsimuleringar som tester. Testerna har genomf¨ortsi en Scania lastbil av anl¨aggningstyp. Experimenten visade p˚aen lateral avvikelse fr˚anrefer- ensbana p˚a7 cm i genomsnitt och en maximal avvikelse p˚a27 cm d˚afordonet k¨ordes p˚a torr asfalt. I den sista delen studeras olinj¨arrespons i fordonets kurvaturreglering och hur detta kan hanteras i MPC’ns prediktions modell av fordonet presenteras ocks˚a.En prediktions modell baserad p˚aen standard kinematisk cykelmodell beskriver inte fordonets laterala r¨orelsetillr¨ackligt bra f¨ordet studerade systemet. Dock, genom att utvidga modellen med en funktion som mappar fordonets respons mot en given kurvaturbeg¨arankan noggrannhet av fordonets r¨orelsef¨orb¨attras.Modellen tillsammans med RA-MPC utv¨arderadesgenom simuleringar och tester. Testerna har genomf¨ortsi en Scania lastbil av anl¨aggningstyp. Utv¨arderingenvisade att den introducerade modellen gav en f¨orb¨attradprecision. Exper- imenten visade p˚aen lateral avvikelse fr˚anreferensbanan p˚a5 cm i genomsnitt och en maximal avvikelse p˚a20 cm d˚afordonet k¨ordesp˚av˚atasfalt. To my parents and brother Acknowledgments There are many who I have to thank for the work presented in this thesis. First, I would like to thank Jonas M˚artensson,for giving me the opportunity to work at the Division of Decision and Control at KTH, for his guidance, and the endless discussions we have. A special thanks goes to one of my co-supervisors Henrik Pettersson, for all his advice and knowledge, and a never ending patience that puts up with me wanting to run just one more experiment. Not to mention the good taste in music that lightens the mood in the vehicles. I am also thankful to Bo Wahlberg, who always grounds me and makes me see the bigger picture giving suggestions and ideas. A big thanks goes to two of my colleagues Pedro Lima and Rui Oliveira. Not only have I collaborated academically with them, but also they are responsible for making my life in Sweden easier and better. A thank you also to Scania and all my colleagues. In particular, I would like to thank Lars Hjorth, Assad Alam, and Kalle Fagerberg for giving me the chance to work at Scania. Moreover, thanks to all the people of the autonomous transport solutions research department that have contributed and developed the vehicle used in the experiments. A special big thanks to the motion planning and control group with whom I had the pleasure of interacting daily. Together with Scania, I would like to thank the Wallenberg Artificial Intelli- gence, Autonomous Systems, and Software Program (WASP) funded by the Knut and Alice Wallenberg Foundation, for financing my project and this research. Furthermore, I would like to acknowledge all my friends and colleagues that have been with me for the past years. From the trips with WASP, to life in Stockholm, and to office work at KTH and Scania, I have always enjoyed the good company, the good mood, and the good discussions. I would like to thank Anamarija, Henrik, Jezdimir, Jonas, Manuel, Oskar, Pedro L., Pedro R., Rui, Sebasti˜ao,and Truls for proofreading parts of this thesis, and Henrik