Control of Quadcopter Considering Collision with Wall Utilizing Monte Carlo Model Predictive Control

Abstract

This paper describes the control of quadcopter that is resistant to collisions with obstacles using Monte Carlo model predictive control (MCMPC). MCMPC is a kind of model predictive control that uses Monte Calro Method to derive the optimal control. MCMPC performs only forward simulations and does not require a gradient of the cost function, hence allowing discontinuous phenomena such as collisions with obstacles to be included in the prediction model. This paper proposes the MCMPC controller that considers impact force by the collision with a wall and minimizes position and attitude errors of a quadcopter when the collision is unavoidable.