¹û¶³´«Ã½

Safety-critical control is the science of designing systems that operate within predefined safe limits. For example, a robot arm in a factory must avoid moving beyond its physical workspace, and a self-driving car must stay clear of obstacles to prevent collisions. These operational boundaries are part of what control theorists call the ‘safe set’. If a system starts and remains within this safe set, it can avoid dangerous scenarios. However, designing controls to guarantee safety in real-world systems, especially those with high complexity and dynamic behavior, is a formidable challenge.

Multiple tasks simultaneously

The difficulty lies in balancing several competing demands. Autonomous systems are often required to perform multiple tasks simultaneously, such as maintaining stability, reacting to changing conditions, and processing information in real time. These tasks are further complicated by limitations like sensor noise, partial understanding of the environment, and the need to operate quickly without sacrificing precision. Traditional methods of ensuring safety, such as rigorous mathematical computations or predefined motion plans, often prove too slow or resource-intensive for systems that need to make split-second decisions.

Balancing computational demands

In recent years, researchers have explored innovative approaches to address these challenges. One promising avenue is the use of Control Barrier Functions (CBFs), which allow systems to manage safety constraints dynamically and in real time. CBFs have shown potential for handling nonlinear systems, which are common in real-world applications. However, even these advanced techniques face hurdles, such as dealing with environmental uncertainty and balancing computational demands with the need for safety and stability.

Significant advancement

This challenge has inspired Ming Li to focus on developing a control synthesis method that bridges theory and practical application. Based on a universal formula he proposes a control design that balances three critical aspects: stability, computation, and safety. This approach adapts to real-world constraints, such as environmental noise and partial observability, while maintaining computational efficiency. By combining advanced theoretical tools with real-world testing, this research demonstrates that autonomous systems can be both safer and more reliable, representing a significant advancement in autonomous control.

Better safety-critical controls

Autonomous technologies continue to evolve and have a direct impact on industries such as transportation, healthcare, and manufacturing. By designing better safety-critical controls, not only the reliability of these technologies is improved, but it also makes them more accessible and trustworthy. Whether it’s enabling safer roads through self-driving cars or creating more efficient factories with advanced robotics, these developments promise to reshape our world for the better—safely and reliably.

Title of PhD thesis: . Promotor: Prof. Siep Weiland. Co-promotor: Assistant Prof. Zhiyong Sun.