Biped gait generation based on parametric excitation by knee-joint actuation

Deterministic Kinodynamic Planning with hardware demonstrations

“…The detail and the derivation of equations (1) and (2) are described in [8]. The dynamics of the biped robot is completely determined by the dynamic equation (1) and the impact equation (2).…”

“…Hence, it is important to design and control relative knee angle appropriately during swing for a robot to walk. To design the control input u, it is sufficient to determine a reference trajectory for relative knee angles of a swing leg, because the existence of the control input completely tracking given relative knee angles was guaranteed by a partial feedback linearization method [8]. We note that, in the developed experimental robot, knee joints are controlled by servomotors.…”

“…Asano et al [7] have first applied this principle to a biped robot having telescopic legs which make the swing leg mass up-and-down by pumping a leg, and have shown that the robot can walk sustainably on a level ground. Later, Harata et al [8] have found that bending and stretching a swing leg has the same effect of pumping a telescopic leg, and have proposed the parametric excitation walking for a kneed biped robot. Harata et al [9] have also proposed the parametric excitation based inverse bending walking in which the knee of swing leg is bent in inverse direction to human movement.…”

“…1, which has a servomotor at each knee joint and has no actuator at a hip joint. In parametric excitation walking proposed by Harata et al [8], [9], not only walking performance but also walking ability strongly depends on the design of reference trajectories for knee joint angles. Furthermore, the reference trajectories are assumed to start from the instance of heel strike, but the developed robot does not have a ground sensor.…”

Minimalistic control of biped walking in rough terrain

“…An important aspect of this model lies in the fact that it is irreducibly simple and analytically tractable, which enable us systematically investigate both mechanical interactions and dynamic behavior control. Previously, the compass gait model was investigated in terms of mechanical interactions in a passive regime (McGeer 1990;Garcia et al 1998;Goswami et al 1998;Su and Dingwell 2007), and its variations were developed for investigating, for example, knee dynamics and locomotion stability (van der Linde 1999; Miyakoshi and Cheng 2004;Asano et al 2007;Harata et al 2007;Kinugasa et al 2008), shapes and actuation of foot segments (Kuo 2002;Ono et al 2004;Adamczyk et al 2006;Kim et al 2007;Kwan and Hubbard 2007), mass distribution (Hass et al 2006), and lateral balancing (Kuo 1999). Control architectures for the compass gait model have also been studied with respect to energy based optimal control (McGeer 1988;Goswami et al 1997;Asano et al 2000;Spong 2003;Spong and Bhatia 2003;Asano et al 2004;Pekarek et al 2007), phase resetting mechanisms and nonlinear oscillators (Kurz and Stergiou 2005;Aoi and Tsuchiya 2005, and control optimization in rough terrains (Pratt et al 2001;Tedrake 2008a, 2008b).…”