a b s t r a c t Conventional models of bipedal walking generally assume rigid body structures, while elastic material properties seem to play an essential role in nature。 On the basis of a novel theoretical model of bipedal walking, this paper investigates a model of biped robot which makes use of minimum control and elastic passive joints inspired from the structures of biological systems。 The model is evaluated in simulation and a physical robotic platform by analyzing the kinematics and ground reaction force。 The experimental results show that, with a proper leg design of passive dynamics and elasticity, an attractor state of human- like walking gait patterns can be achieved through extremely simple control without sensory feedback。 The detailed analysis also explains how the dynamic human-like gait can contribute to adaptive biped walking。85042

1。Introduction

In the fields of biomechanics and robotics, bipedal walking has been investigated for our further understanding of adaptive locomotion mechanisms of human and robots。 Bipedal locomotion in artificial systems was firstly engineered by using predetermined trajectories of the leg joints。 Although this approach demonstrated an outstanding versatility in locomotion behaviors,   adaptivity is highly restricted because this approach requires a precise environment model and demanding computational duty for calculating the trajectories。

Research of passive dynamic walking has questioned the conventional approach。 Based on biomechanical models, the so- called ‘‘compass gait model’’ or ‘‘ballistic walking model’’[1], a number of Passive Dynamic Walkers (PDWs) have been developed and demonstrated natural walking behaviors (e。g。 [2–4])。 Inspired by the muscle activities in the swing leg during human walking, these models utilize no actuation and purely mechanical pendulum dynamics are used to achieve walking behavior。 It has been also shown that, by implementing actuators in PDWs, dynamic walking can be achieved with high energy efficiency and little control even on the level ground [5]。

More recently, a theoretical model, the so-called ‘‘spring– mass walking’’ has been proposed, which demonstrated walking

dynamics with a considerable similarity to that of human [6]。 Although the spring–mass model was originally proposed for running behavior [7–9], an  extension  of  the  original  model  can be applied for walking behavior, which leads to our further understanding of underlying mechanisms of human locomotion。 With a few notable exceptions (e。g。 [10,11]), most of the biped robots with compliant legs were developed for running and hopping behaviors in the past。 Thus the nature of adaptive bipedal walking is still only partially understood。

Based on the theoretical model of bipedal walking with compliant legs, the goal of this paper is to explore a mechanically realistic model of human-like bipedal walking, which can actually be implemented in a robotic platform。 In this paper, we particularly focus on the following two novel features derived from compliant legs: (a) self-stabilizing walking gait patterns with minimal motor control, and (b) joint trajectories resembling that of a human。 Through the experiments in simulation and a real-world robot, we analyze how the elasticity of the leg design can be used for dynamic bipedal locomotion behavior, and compare the behavior of the system with human locomotion。

In the next section, we introduce the walking behavior of a human and the spring–mass walking model。 Then we propose a new locomotion model and test it in simulation and in a robotic platform。

2。Bipedal walking with compliant legs

2。1。Analysis of human walking

In order to understand the influence of compliant legs during walking behavior, we firstly investigate the walking behavior of