Resumen de Symbolic-numeric tools for the analysis of motorcycle dynamics. Development of a virtual rider for motorcycles based on model predictive control
Over the years, the design of modern vehicles is becoming an extremely complex task.
This trend is caused by the increasing functional requirements of new designs and by the vast amount of mechatronic systems within the vehicle. Motorcycle technology has reached a point where, to make further improvements, a global and multidisciplinary understanding of the vehicle is essential. In this respect, it has become indispensable to consider the interaction of such diverse elements as road, suspensions, chassis, engine, rider, sensors and electronics in the design process of new devices.
It is in this challenging framework where simulation takes a crucial role in the evolution and development of new ideas. Virtual models offer a faster and cost-effective alternative to physical testing, they decrease the development time, limit the number of prototypes needed to validate a new design and, most importantly, they permit conducting experiments with no risk for the human operator. This last feature, for instance, results especially attractive for the development of new active safety systems.
In order to achieve realistic results, it is essential to properly replicate the behavior of the human operator within the simulation environment. This is the main scope of the present thesis, which focuses on the application of control techniques to drive a virtual motorcycle model along a predefined path. This implies generating appropriate control actions to govern the throttle, the brakes and the steering system during the simulation.
Two-wheeled vehicles are complex dynamic systems; they are highly nonlinear, non- minimum phase, unstable and underactuated. This makes motorcycle control a rather challenging task. Building upon Model Predictive Control theory, we have developed a novel control strategy capable of riding a nonlinear motorcycle model in a wide range of conditions, including extremely demanding manoeuvres such as those performed by racing riders.
The lack of appropriate simulation packages for vehicle control motivated the in- house development of all necessary tools for the realization of this project. These include, in addition to the virtual rider, a detailed multibody model of the motorcycle and a new methodology for the quasi-steady state analysis of motorcycles.
The research performed in this PhD project yields new insights into the simulation of motorcycle dynamics and extends the working range of existing controllers. Although a sports motorcycle has been used throughout this document, both the modeling approach and the results can be extended to other vehicle typologies. This research has also paved the way for improvements and further developments that are suggested as future research lines.
