MPCC-based Path Following Control for a Quadrotor with Collision Avoidance Guaranteed in Constrained Environments

摘要

This paper describes an autonomous mission on safe traversal of consecutive gates for quadrotors. Considering that the effective sensing range of onboard sensors is practically short, we integrate the path planning into model predictive path following control (MPFC) with sigmoid-based path template which can be easily revised when the new obstacles information is within the detection range. We leverage the framework of model predictive contouring control (MPCC) to suit high speed path following control applications. The MPFC structure allows for input and state constraints for the specific practical scene. The trade-off between contouring accuracy and traversal time is addressed using the MPFC cost function, allowing the system to deviate from the desired path in order to make the path be traversed faster and whereas sacrificing path speed (traversal time) in order to increase safety. The weights in the cost function determine the relative importance of the competing control objectives. In case of traversal, we build a virtual safety corridor in the horizon for quadrotors, which facilitates the control scheme to be extended to incorporate obstacle avoidance by adjusting the constraints and thus re-plan its traversal path. The control performance is investigated experimentally using a realistic dynamics model of DJI M100 platform. The simulation results demonstrate the effectiveness of the proposed control algorithm.