More realistic dynamical modeling of a humanoid robotic arm to reduce control effort with uncertainty using self-impact joint constrained

Abstract

<strong>Background:</strong>The main challenge of modeling humanoid robots is establishing a compromise between the simplicity of the model and accuracy of the system. One of the realities of movement which is important about humans is moving their hands to keep balance and reduce energy consumption while they are walking.<br /> <strong>Methods:</strong> In this context, the role of elbow joint and the limitation that the joint exerts in terms of movement on the forearm and arm as self-impact joint constraint is undeniable. This paper deals with modeling and control of humanoid robot's hand as double-pendulum will consider mentioned constraint while normal walking and also in throwing darts.<br /> <strong>Results: </strong>The presence of the self-impact joint constraint contributed to about a 26% saving in power consumption of robot motors within an impact range of 0.6346 to 0.6896 during normal human walking.<br /> Since this control has a high power, 10 to 30% of the uncertainty was added to the length and mass parameters. As was observed, this controller routed the desired curves in the least possible time.<br /> <strong>Conclusion: </strong>As mentioned earlier, consideration of this constraint in elbow joint of the humanoid robot will help in approaching the reality of system in comparison with past models previously designed. As constraint causes addition of severe nonlinear terms to dynamic system equation, the control of systems with this type of constraint faces a great deal of complexity. For adaptive-neural controller to control of the system of humanoid robot's hand will be used. Also, to display the ability of control system, the uncertainty of length and mass for this system will be considered. The existence of self-impact joint constraint will cause saving in consumption power of robot engines within the impact range.