Objective To address the formation tracking problem of multiple autonomous underwater vehicles (AUVs) under directed asymmetric topology, locally reachable leader information, unknown dynamics, and actuator saturation, a distributed dual-loop event-triggered control method is investigated.
Method The outer loop employs distributed observation, adaptive formation control, and dynamic event-triggered communication to generate desired velocity commands; the inner loop constructs a ten-dimensional augmented model containing command-filtered states, describes input constraints with a non-quadratic cost, and incorporates the filtered input into the HJI (Hamilton-Jacobi-Isaacs) robust envelope; an offline Actor-Critic policy improvement is adopted, and network weights are frozen during closed-loop execution.
Results When the relevant assumptions are satisfied, the outer-loop error satisfies conditional local p-ISS with respect to the inner-loop velocity error, command filtering error, and communication sampling error; the inner loop satisfies local p-ISS and ultimate boundedness; and the communication and control event sequences exhibit no Zeno behavior. Complete dual-loop closed-loop simulation results show that the integral of absolute error (IAE) of position and velocity for the proposed method are 30.34 and 4.51, respectively; the outer loop communicates 1432 times, which is reduced by 71.36% compared with 10 Hz time-triggered communication; the inner loop control updates 10226 times, and compared with a 100 Hz fixed Actor, the position and velocity IAEs are reduced by 0.07% and increased by 3.44%, respectively, while control updates are reduced by 74.44%.
Conclusion The proposed method provides a control scheme that balances tracking performance and communication/control update overhead for multi-AUV cooperative formation under directed topology and unknown dynamics, offering a methodological reference for underwater multi-AUV cooperative operations.
DownLoad: