Model-Free Adaptive Control of Uncertain Autonomous Systems

Model-free adaptive control is a promising approach implemented in various complex and uncertain systems such as unmanned air vehicles, humanoid robots, and autonomous cars. Its key feature is that it does not require an accurate analytical model of the real systems and the dynamic environments since the learning occurs through a trial-and-error process or with partial model knowledge. However, this learning process is time-consuming, computationally expensive, and the obtained solutions are generally not optimal, which may result in failures or poor performances in real-time applications. Therefore, an efficient exploration-exploitation adaptive control strategy is necessary to ease these disadvantages. Under-actuated mechanical systems such as underwater vehicles, robot manipulators, legged robots, quadrotors, and satellites are ubiquitous advanced engineering problems. They have a number of advantages over the fully actuated ones because they weigh less, consume less energy, require simpler communication tools, and fault-tolerant approaches. Furthermore, these under-actuated mechanical systems have fewer actuators to be controlled. Note that it is not possible to control each mechanical component of these systems, and the development of control approaches, especially the model-free intelligent ones, is still a challenging problem. Additionally, external-internal parametric and non-parametric uncertainties such as high-frequency measurement noises, unpredictable chaotic dynamics, control signal time delay, and actuator saturations being unavoidable in real-time applications result in poor control performance.

This Research Topic aims to explore and develop efficient model-free adaptive control strategies for uncertain autonomous systems. The main objectives include addressing the challenges posed by the lack of accurate models, optimizing the learning process to reduce computational costs, and improving the real-time performance of these systems. Specific questions to be answered include: How can we enhance the exploration-exploitation balance in adaptive control? What machine learning techniques can be integrated to improve control accuracy? How can we mitigate the effects of uncertainties and external disturbances?

To gather further insights into the boundaries of model-free adaptive control of uncertain autonomous systems, we welcome articles addressing, but not limited to, the following themes:
- Partially model-free adaptive control approaches
- Fully model-free adaptive control approaches
- Uncertainty prediction and uncertainty control approaches
- Machine learning-based intelligent control approaches
- Vision-based intelligent control approaches
- Data-driven intelligent control approaches
- Adaptive control of under-actuated autonomous systems
- Adaptive control of redundant autonomous systems
- Reduced order observer-based adaptive control approaches
- Reduced order adaptive control of uncertain autonomous systems
- Optimization of the challenging control problems for autonomous systems
- Design, development, and adaptive control of autonomous systems
- Real-time experimental research on adaptive control of autonomous systems