Analytical Criterions for the Generation of Highly Dynamic Gaits for Humanoid Robots: Dynamic Propulsion Criterion and Dynamic Propulsion Potential

Many studies have been made to develop walking anthropomorphic robots able to move in environments well-adapted to human beings and able to cooperate with them. Among the more advanced projects of humanoid robots, one can mention : the Honda robots P2, P3 (Hirai, 1997) (Hirai et al., 1998) and Asimo (Sakagami et al., 2002), the HRP series developed by AIST (Kaneko et al., 1998) (Kajita et al., 2005) (Kaneko et al., 2004) (Morisawa et al., 2005), the small robot QRIO proposed by Sony (Nagasaka et al., 2004), the KHR series developed by KAIST (Kim et al., 2004) (Kim et al., 2005), the last robot of Waseda University having seven degrees of freedom per leg (Ogura et al., 2004), Johnnie (Lohmeier et al., 2004) and H7 (Kagami et al., 2005). These robots are namely able to climb stairs and to carry their own power supply during stable walking. The problem of dynamic locomotion and gait generation for biped robot has been studied theoretically and experimentally with quite different approaches. However, when searchers study the behavior or the design of dynamic walking robots, they inevitably meet a number of intrinsic difficulties related to these kinds of systems : a large number of parameters have to be optimized during the design process or have to be controlled during the locomotion task; the intrinsic stability of walking machines with dynamic behaviors is not robust; the coordination of the legs is a complex task. When human walks, it actively uses its own dynamic effects to ensure its propulsion. Today, some studies exploit the dynamic effects to generate walking gaits of robots. In this research field, four kinds of approaches are used. The first one uses pragmatic rules based on qualitative studies of human walking gaits (Pratt et al., 2001) (Sabourin et al., 2004). The second one focuses on the mechanical design of the robot in order to obtain natural passive dynamic gaits (Collins et al., 2005). The third one deals with theoretical