Time-Varying Trajectory Tracking Formation H_∞ Control for Multiagent Systems With Communication Delays and External Disturbances

Real-time measurement-driven reinforcement learning control approach for uncertain nonlinear systems

“…U (X(ξ), η r (ξ)) dξ employs the optimal strategies η r , ∀r, ν given by (10). This and the stability results yield…”

“…Herein, a gradient-descent approach is employed to adapt the actor and critic weights. Thus, the adaptation error structures of the critic and actor are inspired by the value iteration procedure given by ( 9) and (10), respectively. The tuning error associated with adapting the critic weights is given by E Critic = 1 2 Ŝ (X(t), η(t)) − S (t) 2 , where S (t) is defined by S (t) = U (X(t), η(t)) + Ŝ (X(t + ν), η(t + ν)).…”

“…Hence, many adaptive learning approaches employ either complex or computationally expensive algorithms [5][6][7][8]. This gets more challenging when adaptive mechanisms are adopted for coupled regulation and optimization missions, where the dimensions of the state and action spaces grow significantly [9][10][11].…”

“…This online IRL solution is based on a value iteration mechanism that solves the temporal difference equation ( 9) and updates the control strategy (10) in the above mentioned designated order. Theorem 3 below verifies the convergence of the adapted kernel solution following this iterative procedure.…”

Distributed H∞ consensus of multiagent systems with nonuniform time-varying communication delays

“…In comparison with the study by Ye et al (2014), the leader is dynamic, see equation ( 2). Unlike the studies by Bi et al (2022), Cheng et al (2022), Rezaee et al (2014), and Shariati and Tavakoli (2017), where agents' dynamics are assumed to have no eigenvalues with positive real parts, this article does not impose such a restriction. That is to say, the dynamics can be stable or unstable, which can help reduce the conservatism of the results.…”