Spherical image based visual servoing via nonlinear model predictive control

Abstract

For cameras obeying the unified projection model, a set of independent visual features are designed with a virtual unitary spherical projection process. Then, image based visual servoing is formulated as a nonlinear constrained optimization problem by nonlinear model predictive control in the feature space. Feature jacobian is calculated to define the local model, which is used to predict the evolution of the visual features with respect to the camera velocity over a finite-prediction horizon. Iterative equations for constrained variables about visibility, camera velocity and task space limitation, are designed to meet both 2D and 3D constraints. Finally, simulation results with a classical perspective camera are presented to verify the effectiveness and improved behaviors of proposed method.