Adaptive Control

By C. Hang, T. Lee, W. K. Ho. 1993

Adaptive control is the attempt to " redesign " the controller while online, by looking at its performance and changing its dynamic in an automatic way. Adaptive control is that feedback law that looks at the process and the performance of the controller and reshapes the controller closed loop behavior autonomously. Plant Adaptation Procedure Controller u(t) y(t) y(t) u(t) Figure 1. Adaptive Control as Feedback Law Adaptive Control is made by combining 1. online parameter estimator based on current measurements and control actions 2. control law, that recalculates the controller based on those parameters There are two main modalities of adaptive control. We discussed them below. In this case, the estimated parameters refer to the plant, and the controller is designed based on those parameters. For example, an energy consumption coefficient is estimated; a linearized version of the plant is produced, which is then used in a pole placement method to calculate the gains of the final controller. The main problem is to make sure that the controller behaves well in cases when the estimation of the plant is not good, like for example, during transients. In principle, the indirect method is applicable to all types of plants.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Research Priorities for Robust and Beneficial Artificial Intelligence (2015, background): “Work in adaptive control theory [31], the theory of so-called cyberphysical systems [32], and veriﬁcation of hybrid or robotic systems [33, 34] is highly relevant but also faces the same diﬃculties.”
Gaussian Processes for Data-Efficient Learning in Robotics and Control (2015, background): “Controlling systems under parameter uncertainty has been investigated for decades in robust and adaptive control [4], [35].”
Learning to Adapt in Dynamic, Real-World Environments through Meta-Reinforcement Learning (2018, methodology): “This property has previously been exploited by adaptive control methods (Åström and Wittenmark, 2013; Sastry and Isidori, 1989; Pastor et al., 2011; Meier et al., 2016); but, scaling such methods to complex tasks and nonlinear systems is exceptionally difficult.”
Learning Dexterous Manipulation Policies from Experience and Imitation (2016, background): “Adaptive control on the other hand mostly focuses on learning the parameters of a model with predeﬁned structure, essentially interleaving system identiﬁcation with control ( ˚Astr¨om and Wittenmark 2013).”
Learning to Adapt: Meta-Learning for Model-Based Control (2018, methodology): “This property has previously been exploited by adaptive control methods [3, 42], but scaling such methods to complex tasks and nonlinear models such as deep neural networks is exceptionally difficult, since such models typically require large amounts of data and many gradient steps to learn effectiv”
INDIRECT ADAPTIVE ROBUST CONTROL OF SISO NONLINEAR SYSTEMS IN SEMI-STRICT FEEDBACK FORMS (2002, methodology): “Two approaches have been popular: adaptive control (AC) (Krstic et al., 1995; Landau, 1998) and deterministic robust control (DRC) (Utkin, 1992; Corless and Leitmann, 1981).”
Iterative Value-Aware Model Learning (2018, background): “Some textbooks are Aström and Wittenmark [1994], Krstic et al. [1995], and a recent survey is by Benosman [2018].”
Formulas for Data-Driven Control: Stabilization, Optimality, and Robustness (2019, methodology): “ance requirements in the control design procedure. Literature review Contributions to data-driven control can be traced back to the pioneering work by Ziegler and Nichols [3], direct adaptive control [4] and neural networks [5] theories. Since then, many techniques have been developed under the head”
Research on gain scheduling (2000, background): “This appears, for example, in the chapter on gain scheduling in the "rst edition of A s stroK m and Wittenmark (1995).”
Recent developments on the stability of systems with aperiodic sampling: An overview (2017, methodology): “To the best of our knowledge, this technique was initiated in [172,15], and further developed in [66,254,158] t”