Compliant Leg Design for a Quadruped Robot

malian animals, i.e. that of a cat. It was also influenced by the spring-mass model representation [2]. Compliant prismatic legs show an intrinsic robustness for their angle of attack [2], they have a stability region. If the leg design is changed into a two segmented leg with a specific linear torsional spring, the leg stiffness kleg becomes nonlinear (Fig. 2(b) blue line), similar to that of a declining spring. The influence of the nonlinearity shifts and increases the stability region [3]. We introduce a three segmented, compliant pantograph leg (Fig. 1(c)), its leg stiffness kleg is also non-linear and can be described as piece-wise linear (Fig. 2(b) black lines). Twostep compression springs roughly resemble this behavior. However the nonlinear kleg behavior of our pantograph leg emerges from a linear spring and geometric parameters such as leg segmentation ratio and placement of the spring fixation (d14 in Fig. 1(c)). Choosing a desired leg segmentation ratio during the leg design process changes e.g. the magnitude of the stiffness constant at higher leg compression (Fig. 2(b), solid and dashed black line). If d14 is altered, the point of discontinuity is completely shifted (Fig. 2(b), solid black line and dot-and-dashed black line). Latter can be used to create an Adaptive Compliant Pantograph leg, by on-line changing the switching point e.g. by a motor moving the fixation point of the spring. Often a pre-tension is applied to change spring stiffness, however this only shifts the new working area along the old stiffness trajectory. By e.g. changing the spring fixation point of the pantograph leg (enlarging d14) we can create a completely new stiffness trajectory, and enventually adapt to different load conditions per leg due to a gait change. (a) Cheetah robot 0 10 20 30 40 50 0 5 10 15 20