Adaptive and Data-Driven Innovations in Control Systems

The recent advancements in control systems for autonomous and robotic applications have shown a significant shift towards more adaptive and data-driven approaches. There is a notable emphasis on improving the efficiency and robustness of model predictive control (MPC) schemes, with innovations such as variable-horizon MPC and adaptive terminal constraints leading to reduced conservatism and enhanced performance in tasks like orbital rendezvous. Additionally, the field is witnessing a push towards proactive motion planning methods that avoid local minima in complex environments, leveraging techniques such as repulsive potential augmentation in MPC to achieve global optimality without compromising computational efficiency. Data-driven approaches are also being integrated into trajectory planning for autonomous racing, where velocity prediction models are optimized using Bayesian methods to enhance performance at high-curvature tracks. Nonlinear control systems, particularly in hydraulic actuators, are benefiting from dynamic input mapping inversion to avoid algebraic loops and improve tracking performance. Lastly, there is a growing interest in developing unitless and normalized measures of nonlinearity for state estimation, which can adaptively select or parameterize estimation algorithms based on the system's nonlinearity.

Sources

Robust Variable-Horizon MPC with Adaptive Terminal Constraints

Towards Local Minima-free Robotic Navigation: Model Predictive Path Integral Control via Repulsive Potential Augmentation

A Data-Driven Aggressive Autonomous Racing Framework Utilizing Local Trajectory Planning with Velocity Prediction

RTI-NMPC for Control of Autonomous Vehicles Using Implicit Discretization Methods

Dynamic Input Mapping Inversion for Algebraic Loop-Free Control in Hydraulic Actuators

Design of Unitless Normalized Measure of Nonlinearity for State Estimation

Built with on top of