Decentralized control functions in trajectory guidance of a non-holonomic AUV

This paper describes a modification of authors´ previous work on virtual potential methods for planning and guidance of autonomous underwater vehicles (AUVs) along trajectories in ℝ² and ℝ³. The modification replaces the algebraic sampling procedure of scalar potentials, performed in the previous algorithm in order to allow for the numerical approximation of the true local gradient of the potential, with the direct analytical solution for the local gradient. The modification also allows for a far more flexible integration of rotors, allowing fine-tuning by revealing analytical relationships between the stator and rotor components of calculated acceleration vectors. The modification is critical to the control of holonomically constrained AUVs with finite dynamics and significant lags in response to rudder and change of propeller rpms.