A Culling Procedure for Collision Avoidance Model Predictive Control with Application to Ship Autopilot Models

An optimization-based approach for collision avoidance of fully-dimensional autonomous vehicles in confined environments is considered. This is realized by means of a dual reformulation of the signed distance function which yields an exact representation of the usually nonlinear, non-convex, and non-differentiable signed distance function. However, the dual collision avoidance MPC induces additional decision variables which greatly increase problem complexity. Therefore, a culling procedure is combined with the dual approach in order to reduce the problem size by identifying and eliminating decision variables associated with faces of the polyhedral obstacles and controlled vehicle, respectively. The proposed method is illustrated using a simulative study where a path-following task is performed for a ship autopilot model and near-field obstacle information is extracted from a set of AIS data from the Kiel bay area. Copyright (C) 2021 The Authors.