Original title: On the Degree of Safety: Beyond Safe or Unsafe with Control Barrier Functions
Ruoyu Lin, Fabio Pasqualetti, Magnus Egerstedt · September 2026
Control barrier functions are mathematical tools that tell an automated system whether it is safe or unsafe. But they only give a simple yes-or-no answer. They do not measure how safe the system actually is. Researchers checked whether values, slopes, or solvable formulas from these functions could measure safety depth. They found that these math numbers can be altered without changing the safe zone, so they do not show true safety levels. A formula might also fail either from real danger or just poor math design. To fix this, the researchers made a new metric called invariance authority demand. It measures the control effort needed to keep the system safe, helping engineers repair unsafe designs or improve their motors.
Imagine you're packing for a trip and checking if your bag is under the weight limit. A simple scale says yes or no, but it doesn't tell you how close you are to the limit or what happens if you add one more item. In this paper, the current safety tool is like that scale—it only says safe or unsafe. The authors' new measure is like a tool that shows exactly how much weight you can add before the bag breaks, giving a real sense of margin.
Engineers designing self-driving cars or robots rely on safety checks that only say safe or unsafe. This paper shows those checks can be misleading and offers a better way to measure how much control is needed to stay safe, which could prevent accidents by showing when systems are close to failure.
The abstract doesn't mention experimental results or real-world testing. It focuses on theoretical analysis and introduces IAD, but doesn't show how it works in practice or compare it to existing methods.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
On the Degree of Safety: Beyond Safe or Unsafe with Control Barrier Functions — Ruoyu Lin, Fabio Pasqualetti, Magnus Egerstedt
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.