Welfare – Safety

Safety is one of the most immediate and widely recognised concerns in human-robot interaction. Unlike more abstract values, it tends to surface early in both design discussions and user expectations, because it connects directly to physical integrity and perceived risk. Whether robots are used in care settings, homes, or public environments, users tend to ask a simple question: can this system cause harm, or prevent it? This makes safety a baseline condition for acceptance, rather than an optional feature.

Safety in HRI is typically framed as the requirement that robots should not pose a risk of physical injury to users, including vulnerable populations such as older adults or children. This includes constraints on force, movement, and system behaviour, as well as design choices that reduce the likelihood of accidents. Beyond direct physical harm, safety also extends to the ability of robots to monitor environments and respond to dangerous situations in a way that supports wellbeing. For example, systems may be designed to detect falls, trigger alerts, or intervene in potentially hazardous scenarios. At the same time, this introduces a tension between intervention and user autonomy, particularly when safety measures involve surveillance or external notification.

As noted in the focus group findings, safety is not only about preventing immediate harm, but also about ensuring that robots can be used without creating new forms of vulnerability or discomfort, especially in everyday domestic contexts. This concern is strongly reflected in the literature, often in very concrete terms. Safety is described as the ability to avoid pain or harm, reinforcing its role as a fundamental requirement in companion systems. In care contexts, robots are envisioned as systems that can detect falls and trigger alarms to support timely intervention, or even anticipate dangerous situations and take preventive actions such as turning off appliances or restraining movement when necessary. These capabilities are also tied to broader notions of security and continuous monitoring, where robots contribute to perceived safety through constant presence and alert systems. In dementia care, safety is explicitly positioned as a primary ethical goal alongside wellbeing and harm prevention. Empirical studies with older adults similarly report that robots can increase both physical and psychological feelings of safety, for instance by monitoring users and notifying caregivers in case of emergencies.

At the same time, real-world deployments have also revealed the importance of robust engineering, as accidents such as entrapment or collisions have occurred in field settings. Finally, design guidelines often emphasise limiting physical force and ensuring smooth, predictable motion to minimise risk, especially when robots operate near children or other vulnerable users.