Safe Care

What it is

Safe Care is an integrated architecture for AI-supported elderly-support robots. A deterministic Safety Constraint Gate sits between proposal and action: the language model proposes; the gate decides what may execute. It is aimed at care organisations, research partners, and clinical stakeholders as an information architecture — not as a claim of a certified medical device.

How it relates to Volant

Volant remains the operator layer for mixed industrial fleets on the ground and in the air. Swarming holds the SWARMind air-coordination lineage. Safe Care addresses a different setting: ageing in place and fragmented care robotics. All three sit under HOPn; they do not replace each other.

Five coordinated layers

1. 1

   Perception

2. 2

   Digital twin and risk (Resident Care Digital Twin)

3. 3

   Constrained reasoning — language model plus Safety Constraint Gate

4. 4

   Action and human–robot interaction

5. 5

   Care-network interfaces

Design principle

The language model proposes; the Safety Constraint Gate disposes. Policy enforcement is meant to be deterministic, so unsafe suggestions are blocked rather than negotiated away in chat.

Designed to support care teams

Copy and product intent are conservative: Safe Care is designed to support care teams and family decision-makers with constrained assistance. It is not positioned as a replacement for care workers, clinicians, or human oversight.

Evaluation scope

The Safe Care project reports module- and simulation-level evaluation (including wearable fall-detection benchmarks and gate stress tests described in that project). No living-lab clinical validation study has been performed. Nothing on this page is a medical, diagnostic, or certification claim.