Control Methods for Bipedal Walking Robots With Integrated Elastic Elements

Introduction

Walking robots are one of the central research topics in mobile robotics, due their wide applicability and their potential for natural integration into social and industrial processes, allowing automating a variety of labor-intensive tasks (Luk et al., 2006; Armada et al., 2005; Buschmann 2010). Particular applications for legged robots range from disaster response and to pipe inspection (Bouyarmane et al., 2012; Savin & Vorochaeva, 2017a).

There had been a lot of research focused on particular types of walking robots, including bipedal robots, quadrupeds and multi-legged robots (Hurmuzlu et al., 2004). Bipedal robots present particular interest due to possibility of using anthropomorphic designs and implementing anthropomorphic motions, which can be beneficial in terms of integrating of this type of robots into the existing processes and infrastructure (Ogura et al., 2006, Khatib et al., 2004). Examples of anthropomorphic biped designs can be found in (Kaneko et al., 2008; Kaneko et al., 2011; Khusainov et al., 2015; Hirose et al., 2007). Non-anthropomorphic bipeds can be found in (Xie et al., 2018; Clary et al., 2018). A separate class of humanoid walkers are active exoskeletons, which are inherently anthropomorphic (Yan et al., 2015; Young et al., 2017; Panovko et al., 2016; Jatsun et al., 2018).

One of the benchmarks for the walking robot technology had been DARPA Robotics Challenge (Johnson et al., 2015; Feng et al., 2015), which demonstrated of the progress in the bipedal robotics and simultaneously revealed a number of problems that still need to be addressed. One of these problems is safe interaction between the robot and the environment, including obstacles, tools, people and other robots. Currently used practical solutions to these challenges are to minimize the robot’s interactions with the environment as much as possible, and to act as a stationary manipulator when a manipulation task is present (Atkeson et al., 2015). This leads to stiff and inefficient motions and non-robust control schemes. The capability to operate safely in the scenarios where unplanned contact and collisions are inherently possible has not yet been achieved by humanoid robots.

One of the ways this capability can be achieved is by designing the robots to be less rigid (Pratt, 2002; Robinson et al., 1999). This might allow avoiding damage both to the robot and to the environment in the events of the unplanned contact and collisions. Introduction of elastic elements can also be used for improvement of robots’ energy efficiency in particular regimes of operation (Pratt et al., 1997). However, this type of robot design requires changes in the control strategies employed for bipedal robots.

This chapter looks at the problem of controlling bipedal walking robots with integrated elastic elements. It is a broad class of robots with varying structures. This problem can be partially solved by using existing techniques developed for walking robots with rigid structures. This is true for the algorithms based on the robot’s kinematics and geometry, rather than their dynamics. However, some of the elements of the control pipeline, including feedback controller design and controller parameters tuning require separate analysis.