Section: Systems and Control
Tutor: GARATTI SIMONE
Advisor: SAVARESI SERGIO MATTEO Major Research topic
:Force-Feedback Control Systems Design for High Performance VehiclesAbstract:
Automotive scenario is constantly evolving, pursuing more and more challenging problems and spreading among many different engineering fields. From an industrial perspective, performance and safety are requirements pushing towards technological improvements, especially concerning the so called ‘Drive-By-Wire’ framework, which branches off in many different aspects, from braking to steering actuators, and more.Moreover, in the recent years, almost all the manufacturers are leading the market into an electrical “revolution”, starting from hybrid vehicles, which represents the main alternative to standard ICE vehicles. This is also paving the way to new frontiers in the research field, being electric cars more suitable and flexible than standard ones. As an example of this concept, in-wheel electric motors represent a future outstanding solution to enhance performance and flexibility, especially in the control framework, being possible to modulate traction and braking force at each wheel independently.The common link of all those development directions is the framework related to “autonomous vehicles”, which merges together all the technological challenges arising in automotive, and not only.Being immerged in such scenario and aiming to bring a contribution inside this wide research world, an interesting question arises: “Which variables have not already, totally or partially, been exploited in vehicle control?”. Well, undoubtedly “Forces”, which in turn represent the topic investigated in this PhD project. Practically, I consider the use of forces inside two different layers of the control architecture of a vehicle.
The internal layer concerns the lower-level part of the control scheme, namely the braking actuation unit. I decided to investigate one the most interesting and promising technology in this framework, namely an electro-mechanical Brake-By-Wire Actuator. This architecture has gained lot of attention, representing the new frontier in terms of achievable performance, evaluated both in term of braking power and reliability. These actuators have no hydraulic components and thus the braking force exerted between the pads and disk is the only variable which matters to characterize the braking maneuver. As a consequence, the focus will be related to clamping force control problem, addressing both estimation and controller tuning issues, in order to guarantee also proper actuation capabilities for higher-level controllers. Moving one step above in the control architecture, the main focus of the research project lies in the study of tire-road contact forces. They are well-known to be fundamental in the vehicle dynamics, governing the motion of a vehicle and defining almost completely its performance and stability.In classical vehicle stability control systems, their knowledge is not employed at least directly, even due to the absence of commercial reliable sensors. The main goal of this research is to answer to the following question “How and where the knowledge of tire-road forces could be employed to enhance control performance?”. As for the internal layer, two different problems could be addressed in this field. The first one is directly linked with the previous question, namely how the information about tire-contact forces could be integrated in a vehicle control system, aiming to improve performance in complex and uncertain driving scenarios. On the other side, also in this framework the employment of force sensors is far from being an applicable solution on real commercial vehicles. As a matter of fact, both in-tire sensors and dynamometric wheels have huge limitations, respectively in terms of bandwidth/accuracy and dimensions, so that the employment of virtual sensors represents one of the most viable and cost-effective solution. An important part of this research project is related to this issue, addressing the problem from two different perspectives, namely model-based and black-box.