Spring-Mass Jumping of Underactuated Biped Robots

Abstract

In many types of fast animal and human locomotion an almost sinusoidal pattern is observed for the ground reaction force; therefore, a simple spring-mass model can approximate the generally observed force pattern. The adjustment of the leg during running, jumping or hopping is addressed using a spring-mass model with a fixed landing angle of attack with the objective of obtaining periodic movement patterns. We found that this self-stabilizing spring-mass model can be applied as a movement criterion for biped joints’ trajectory generation in jumping. To create desired velocity and stride-to-stride length, a synchronization method was applied between biped nonlinear dynamics and spring-mass dynamics as slave and master dynamics, respectively. The results of performed simulations show that while our model lacks the flight phase and impact model for a complete cycle of jumping process, this technique might become of great use in the future’s biped path planning which we call it “dynamic path planning”.