Content area
Full Text
1. Introduction
In recent years, many significant achievements have been made on the research of biped robots, which has gained an increasing interest. A lot of biped robots such as ASIMO (Hirai, 1999), HUBO (Park et al., 2007) and HRP (Kaneko et al., 2002) can walk on the flat terrain stably, but unnaturally and without consideration of energy consumption consumed by all actuation motors, which reduces the operation time of the battery largely (Kuo, 2007). They may fall down sometimes during walking on the non-flat terrain. Hence, it is necessary to make appropriate tradeoff between the walking stability margin and the energy consumption according to the conditions of terrain for helping biped robots to more effectively adapt to different terrains, as shown in Figure 1. So, effective gait planning and control strategy for biped robots waking through different terrains, which considers the relationship between stability and energy efficiency fully, is indispensable.
The studies on gait planning and control scheme for biped robots can be classified into four main groups:
modeling of biped robot;
online gait motion synthesis (GMS) main including trajectories generation and inverse kinematics;
gait parameter optimization (GPO) according to the appropriate performance index; and
dynamical balance control scheme.
In the first research group on modeling of biped robots, many researchers have introduced several simplified models instead of accurate model with consideration of the complexity of computation and control. Kajita proposed a simplified model three-dimensional (3-D) linear inverted pendulum mode (3-D LIPM), which is used to generate the trajectory of the center of mass (CoM) with constant height for biped robots easily (Kajita et al., 2001). Park and Kim (1998) improved 3-D LIPM using gravity compensation with consideration of swing foot mass. Albert and Gerth (2003) proposed a two-mass IPM with also consideration of swing foot mass. With consideration of the mass of two feet, Sato et al. (2011) proposed a three-mass IPM and Shimmyo et al. (2013) extended it to preview control.
For the research group on online GMS, many researches have adopted zero moment point (ZMP) proposed by Vukobratovic and Borovac (2004) and 3-D LIMP or three-mass IPM to generate CoM reference trajectory of a biped robot in real time (Kajita et al., 2002; Shibuya et al.,...