Feedforward and H_inf feedback robotic force control in a 1-dof physical interaction using a nonlinear human model

Mourad Benoussaad (1) , Micky Rakotondrabe (1)

1 LGP - Laboratoire Génie de Production

Robot interacting with exible object, and particularly with human in movement, is an increasing topic in several applications of today. It is particularly interesting and challenging in term of modeling and control because of the human variability, uncertainty and non-linearity. In the current work, we propose an original approach to model and control the interaction between a 1-dof robotic system and a human. While the human model includes a nonlinear biomechanical model, the nonlinearity is accounted and compensated for in a feedforward controller. Then a robust H1 feedback control is added in order to allow the closed-loop satisfy some specified performances under possible uncertainties and under external disturbance such as noise. The feedforward-feedback control is afterwards extensively verified in simulation which confirm its efficiency.

Mots clés: Human-robotic interaction; Modeling; Nonlinear feedforward control; H1 feedback control; 1-dof

Domaines: Robotique [cs.RO]

Mourad Benoussaad, Micky Rakotondrabe. Feedforward and H_inf feedback robotic force control in a 1-dof physical interaction using a nonlinear human model. IFAC - World Congress 2020, Jul 2020, Berlin (on line), Germany. ⟨10.1016/j.ifacol.2020.12.1415⟩. ⟨hal-03242687⟩