No description available.

With the rapid expansion of the range of digital characters involved in film and game production, creating a wide variety of expressive characters has become a problem that cannot be solved efficiently through current animation methods. Key-frame animation is time-consuming and requires animation expertise. Motion capture is constrained by equipment and environment requirements and is most applicable to humanoid characters. Simulation can produce physically correct motion but does not account for expressiveness. This thesis focuses on developing a more efficient animation system using a procedural approach in which the skeletal structure and characteristics of motion that communicate weight and age in quadrupeds have been isolated and engineered as user-controlled tools and modifiers to build creature shape and synthesize cyclic gait animation. This new approach accomplished the goal of quick generation of expressive characters. It is also successful in achieving real-time animation playback and adjustment.

Bipedal locomotion is a challenging control engineering problem due to the non-linear dynamics and postural instability of the bipedal form. In addition to these challenges, some dynamical effects such as the ground reaction force are difficult to model accurately in simulation. To this end, it is essential to develop physical hardware to validate walking control strategies and gait generation methods. This thesis develops an on-line walking control strategy for humanoid robots and the electromechanical design of a physical platform for experimental validation. The first part of the thesis presents the development of a 14 degrees-of-freedom (DOF) lower body humanoid robot. The initial electromechanical design of the proposed system is derived from dynamic modeling of a general multibody system. Kinematic trajectories for the lower body joints are extracted from motion captured human gait data to form the preliminary design specifications. The drivetrain components are selected by analyzing the mechanical power requirements, torque-speed profiles, efficiency and thermal characteristics of actuators. The supporting mechanical chassis and power transmission system are designed to raise the center-of-mass (to reduce the swinging inertia of each leg) while minimizing the overall weight of the system. Refining the design of a complex multibody robotic system like the biped is an iterative process. The mechanical model of the system is transferred from Computer-Aided-Design (CAD) software to a dynamic simulator for analysis and the design is revised to improve performance. This iterative approach is necessary as small changes in the mechanical model can have significant impact on the overall dynamics of the system as well as implications for control design. A streamlined prototyping toolchain is developed in this thesis to extract the relevant kinematic/dynamic parameters of a mechanical system in CAD and automatically generate the equivalent system in a dynamic simulator. This toolchain is used to revise the electromechanical design and generate forward dynamics simulations. The second portion of this thesis develops a novel walking control strategy for on-line gait synthesis for 3D bipedal robots based on Wight's Foot Placement Estimator (FPE) algorithm. This algorithm is used to determine the desired swing foot position on the ground to \emph{restore} balance for a 2D bipedal robot. The FPE algorithm is extended to the general 3D case by selecting a suitable plane in the desired direction of motion. Complete gait cycles are formed by combining a finite state machine with the 2D FPE solution along the selected plane. Gait initiation is accomplished by computing state-dependent task space trajectories on-line to produce a forward momentum along the selected plane. A whole-body motion control framework (Jacobian-based prioritized task space control scheme) tracks the task space trajectories and generates the appropriate joint level command for each state. The joint level commands are tracked by local high gain PD controllers. This framework produces the desired whole-body motion during each state while satisfying higher priority constraints. Gait termination is accomplished by controlling the swing foot position to track the FPE point on the ground along the selected plane. The proposed control strategy is verified in simulation and experiments. A parallel hardware-in-the-loop (HIL) testing environment is developed for the physical lower body humanoid robot. The motion control framework and joint dynamics used in the proposed walking control strategy are verified through HIL experiments.

Pure reinforcement learning does not scale well to domains with many degrees of freedom and particularly to continuous domains. In this thesis, we introduce a hybrid method in which a symbolic planner constructs all approximate solution to a control problem.. Subsequently, a numerical optimisation algorithm is used to refine the qualitative plan into an operational policy. The method is demonstrated on the problem of learning a stable walking gait for a bipedal robot. The contributions of this thesis are as follows. Firstly, the thesis proposes a novel way to generate gait patterns by using a genetic algorithm to generate walking gaits for a humanoid robot using zero moment point as the stability criterion. This is validated on physical robot. Second, we propose an innovative generic learning method that utilises the trainer's domain knowledge about the task to accelerate learning and extend the capabilities of the learning algorithm. The proposed method, which takes advantage of domain knowledge and combines symbolic planning and learning to accelerate and reduce the search space of the learning problem, is tested on a bipedal humanoid robot learning to walk. Finally, it is shown that the extended capability of the learning algorithm handles high complexity learning tasks in the physical world with experimental verification on a physical robot.

vii, 343 leaves : diags. tables. ; 30 cm / Title page, contents and abstract only. The complete thesis in print form is available from the University Library. / Thesis (Ph.D.)--University of Adelaide, Dept. of Statistics, 1978

Thesis (Ph. D.)--Ohio State University, 2007. / Title from first page of PDF file. Includes bibliographical references (p. 184-194).

Lic.-avh. (sammanfattning) Stockholm : Karolinska institutet, 2007. / Härtill 2 uppsatser.

Thesis (PH.D.)--Columbia University, 1929. / Bibliography: p. 231-232.

Thesis (Ph. D.)--University of Missouri-Columbia, 1997. / Typescript. Vita. Includes bibliographical references (leaves 291-307). Also available on the Internet.

Thesis (Ph. D.)--University of Missouri-Columbia, 1997. / Typescript. Vita. Includes bibliographical references (leaves 291-307). Also available on the Internet.