Currently, safety is predominantly considered on a risk basis and treated probabilistically, including serious incidents such as fatalities. In contrast, this study examines safety from a logical perspective, termed safety logic. Logical approaches to safety have historically been applied only in limited fields, such as railway signal systems and logic circuits, and have rarely been generalized to broader safety design. This study proposes a logical framework to predict and avoid outcomes that cannot be directly experienced, using deductive reasoning. For example, given the proposition “Operation has the potential to cause incidents,” its contrapositive, “Stopping operation before an incident prevents experiencing an incident,” also holds. From this, it is logically required to stop before an incident occurs. An incident is represented using logical formulas as p∧q=0, leading to two key conditions: stopping the cause (operation) (¬p≥q) and avoiding being a victim regardless of operation (¬q≥p). The latter is defined as the risk formula, enabling risk to be expressed in logical form. We propose that machines, as causal elements, be designed so that operation is inherently conditioned on stopping prior to incidents, thereby integrating logical reasoning into safety design.
Murakami et al. (Wed,) studied this question.