Analysis, Control and Design of Walking Robots

In this thesis five research questions are discussed that are related to the development of two-legged (bipedal) walking robots. The research questions are categorized in three main topics: analysis, control and actuation and design. The research questions are: - How can we analyze the behavior of a 2D passive dynamic walker that is walking on rough terrain? -- An extension to the Poincare̿ mapping is proposed that allows the use of this method on non-flat terrain. - By looking at the robot from a different 'perspective', can we gain more insight in its dynamics? -- The use of coordinate transformations is discussed, a coordinate-free interpretation of the Zero-Moment Point is given and an analysis of a simplified model of the robot (the 'locked inertia model') is given. - How can we control a walking robot in order to stabilize it in the lateral (sideways) direction? -- A linear controller which takes advantage of some specific properties of the robot model is introduced, and experimental results of the 'extrapolated center of mass' are discussed. - How can we improve the actuators in order to get minimum energy consumption? -- A concept is introduced for a new type of actuator which can store negative work mechanically and re-use it later. - How can we improve the knee and ankle joints of a walking robot? -- New mechanical designs for the knees of the 2D walking robot Dribbel and the ankles of the 3D walking robot TUlip are proposed. Each chapter of this thesis is based on an article which was published at or submitted to a conference.