Catálogo de publicaciones - libros

This is a book about biped robots. Why would someone research biped robots? In ancient Dutch, the answer would be: ‘Ter leering ende vermaeck’, which means ‘for learning and for fun’. From the research, we can learn the principles of dynamic two-legged walking. This knowledge can be used for biomechanics (understanding human locomotion), prosthetics (aiding human locomotion), and robotics (building human-like robots for entertainment and for assistance tasks).

The concept of passive dynamic walking is the starting point for our research. In this chapter, we would like to review the work that has been done on passive dynamic walking. Although we have made some contributions of our own to this field, most of the research reported in this chapter was done by others.

This chapter reports the development of the pneumatic components that we used for all robots in this book. The choice for pneumatic actuation has advantages as well as drawbacks compared to actuation with electric DC motors. The advantage is the inherent compliance of the McKibben muscles (Section 3.1). The good match between the compliant McKibben muscles and the concept of passive dynamic walking was the main argument in favor of the choice for pneumatics. The main drawback is the lack of sufficiently lightweight commercially available components, which means that many of the components must be specially developed for the research project (Section 3.2). A second drawback is the difficulty of implementing high-bandwidth control, which may be a reason for us to switch to electric DC motors for our future prototypes. However, for the phasic (i.e. once-per-step) actuation that we have implemented in all robots in this book, the pneumatic system is a satisfactory choice (Section 3.3). The complete, autonomous pneumatic system can successfully power a biped robot as demonstrated with the prototype ‘Baps’ (Section 3.4). The content of this chapter is based on four of our papers [157, 159, 158, 186].

Purely passive walkers (Chapter 2) and our first prototype Baps (Chapter 3) can walk efficiently, but they fall too easily. Now it is time to do something about it. The tools are here: we now have knowledge of passive dynamic walking and we have a way of actuating these walkers with pneumatic McKibben muscles. How can the actuation be used to enhance the stability of the walking motion? This chapter will provide an answer, first by analyzing the stability of an unactuated model, followed by an implementation of hip actuation both in an elementary simulation model and in the prototype ‘Mike’ (Fig. 4.1) that was built for this purpose. The content of this chapter is based on two of our papers [141, 184], © 2005 IEEE, reprinted with permission.

More than two thirds of the mass of a human being is located in the upper body (head-arms-trunk), while the legs are relatively lightweight. The addition of an upper body to passive dynamic walkers remains an active research topic; so far only the legs have been considered. The problem is that the upper body should be stabilized in the upright position, while at the same time the alternating swing leg should be able to swing passively to a forward position. Some passive solutions have been found [17, 58], in which the upper body is another passive pendulum-like component. Also a number of active control solutions have been proposed, such as McGeer’s ‘levered isotonic tendons’ [110], variable springs [160] or a controllable ‘backlash clutch’ in the hip joints [122], all fairly complex solutions. In contrast, we have searched for an alternative, mechanical solution.

One of the grand challenges in biped walking is to find the key to stability in three dimensions. In addition to the fore-aft motions (pitch), in 3D also sideways motions (lean) and rotations around the vertical axis (yaw) are possible. It is the interaction between all three of these that makes the problem of 3D stability so difficult. Consequently, most of the known solutions to this problem are successful because they in one way or another:

In this chapter, however, we aim not at a between the degrees of freedom, but conversely we show how a between lean and yaw can actually improve the stability of the walking motion.

The purpose of this book is to present our work on pneumatic biped robots, so that you may build upon it and avoid our mistakes. Thus, a quintessential part of the information that this book should transmit are our afterthoughts; things that we haven’t done (yet) but that we have developed a vision on. These things are presented in this chapter. We hope that these insights will prove useful for future research.