https://www.mdu.se/

Autonomous control of mobile robots is a research topic that has received a lot of interest. There are several challenging problems associated with autonomous mobile robots, including low-level control, localisation, and navigation. Most research in the past has focused on developing algorithms for three or four-wheeled mobile robots, such as autonomous cars and differential drive robots, which are statically stable systems. In this thesis, autonomous two-wheeled robots are considered, such as autonomous bicycles, which are naturally unstable systems, and without proper actuation, they will lose balance and fall over. Thus, before developing algorithms for higher-level functionality such as localisation and navigation of an autonomous bicycle, the balance of the bicycle needs to be addressed. This is an interesting research problem as the bicycle is a statically unstable system that has proven difficult to control, but given adequate forward velocity, it is possible to balance a bicycle using only steering actuation. Moreover, given a sufficient forward velocity, the bicycle can even become self-stabilised.

In this thesis, the balance and trajectory tracking of an autonomous bicycle is investigated. First, we propose an extension of previously proposed bicycle models to capture the steering dynamics including the motor used for controlling the handlebar. Next, several control methods which can stabilise an autonomous bicycle by actuation of the steering axis and the forward velocity of the bicycle are developed. The controllers are compared in simulations on both a linear and nonlinear bicycle model. The simulation evaluation proceeds with experiments conducted on an instrumented bicycle running on a bicycle roller. Moreover, trajectory tracking of an autonomous bicycle is addressed using a model predictive controller approach where the reference lean angle is computed at every sample interval and is tracked by the balance controller in the inner loop. Finally, path planning in a static environment is considered where the proposed strategy realises a smooth path that adheres to the kinematic and dynamic constraints of the bicycle while avoiding obstacles and optimises the number of heading changes and the path distance. The results obtained from detailed multibody simulations highlight the feasibility of the balance controller, trajectory tracking controller, and path planner. 

Autonomous bicycles constitute challenging benchmark systems for control, due to their nonlinear, non-holonomic, and in general, underactuated, open-loop unstable dynamics. Traditional model-based controllers such as proportional-integral-derivative (PID) controllers and linear quadratic regulators (LQRs) can stabilize the bicycle, but rely on simplified models that may not capture unmodelled and time-varying effects. In contrast, recent direct data-driven control methods based on Willems’ fundamental lemma bypass explicit modelling, yet typically assume linear time-invariant dynamic systems and require persistently exciting inputs that are difficult to apply safely on unstable systems. 

This thesis investigates how traditional model-based and direct data-driven control methods can be used, and combined, to balance and guide an autonomous bicycle using mainly steering actuation as input. First, PID, LQR, fuzzy controller, feedback linearization (FL), and direct data-driven controllers are designed and compared in high-fidelity simulations and experiments on an autonomous bicycle. The results show that classical model-based controllers provide strong baselines, while direct data-driven controllers can enhance performance when combined with classical controllers. Second, a unified framework is proposed in which an inner-loop FL controller stabilizes and partially linearizes the bicycle, and an outer-loop direct data-driven controller operates on the FL-stabilized system. Two different types of direct data-driven methods are evaluated in this setting: a static, nonlinear controller and the Data-enabled Policy Optimization (DeePO) algorithm. Third, the DeePO algorithm is analysed and modified to mitigate state perturbations, leading to a perturbation-free variant studied on LTI systems. Finally, a model-based PID–MPC trajectory tracking scheme is compared with a data-driven framework relying on Data-enabled Predictive Control (DeePC) for trajectory tracking, combined in a cascade architecture with the FL-DeePO setup. Simulations show that while PID–MPC achieves better tracking accuracy, the data-driven cascade attains successful trajectory tracking without relying on an explicit dynamic model.

Autonoma cyklar utgör utmanande riktmärkessystem för reglering, på grund av sin icke-linjära, icke-holonoma och i allmänhet underaktuerade, öppen-slinga-instabila dynamik. Traditionella modellbaserade regulatorer, såsom proportio­nal–integral–derivata-regulatorer (PID) och linjärt kvadratiska regulatorer (LQR), kan stabilisera cykeln, men bygger på förenklade modeller som inte nödvändigtvis fångar omodellerade och tidsvarierande effekter. I kontrast till detta kringgår nyare direkta data-drivna regleringsmetoder baserade på Willems fundamentallemma explicit modellering, men antar typiskt linjära tidsinvarianta dynamiska system och kräver ständigt exciterande insignaler som är svåra att tillämpa på ett säkert sätt på instabila system.

Denna avhandling undersöker hur traditionella modellbaserade och direkta data-drivna regleringsmetoder kan användas, och kombineras, för att balansera och guida en autonom cykel med huvudsakligen styraktuering som indata. För det första konstrueras PID-, LQR-, fuzzyregulatorer, feedbacklineariseringsregulatorer (FL) samt direkta data-drivna regulatorer, vilka jämförs i högfidelitets-simuleringar och experiment på en autonom cykel. Resultaten visar att klassiska modellbaserade regulatorer utgör starka riktmärken, medan direkta data-drivna regulatorer kan förbättra prestandan när de kombineras med klassiska regulatorer.

För det andra föreslås ett enhetligt ramverk där en innerloop-FL-regulator stabiliserar och delvis lineariserar cykeln, medan en ytterloop-regulator baserad på direkt data-driven reglering verkar på det FL-stabiliserade systemet. Två olika typer av direkta data-drivna metoder utvärderas i denna konfiguration: en statisk, icke-linjär regulator och algoritmen Data-enabled Policy Optimization (DeePO). För det tredje analyseras DeePO-algoritmen och modifieras för att motverka tillståndsstörningar, vilket leder till en perturbationsfri variant som studeras på linjära tidsinvarianta (LTI) system.

Slutligen jämförs ett modellbaserat PID–MPC-upplägg för banföljning med ett data-drivet ramverk som bygger på Data-enabled Predictive Control (DeePC) för banföljning, kombinerat i en kaskadarkitektur med FL–DeePO-strukturen. Simuleringar visar att PID–MPC uppnår bättre uppföljningsnoggrannhet, medan den data-drivna kaskaden uppnår lyckad banföljning utan att förlita sig på en explicit dynamisk modell.