Social robotics is increasingly used in healthcare and wellbeing scenarios,
addressing various care needs. Evaluating these robots on a large scale
involves challenges, such as transporting and deploying physical robots and
adapting their behavior in the field. This paper proposes a web-based
infrastructure for evaluating SHARA, a social robot aiding elderly people. This
system helps developers and stakeholders overcome evaluation limitations by
using a semi-automatic Wizard of Oz (WoZ) approach. The WoZ operator can
observe the simulated robot behavior and intervene if necessary, enabling a
flexible human-inthe-loop control. The infrastructure streamlines the iterative
evaluation and improvement of care services, allowing simulation-based feedback
on robot interfaces and user interactions before physical implementation. This
approach reduces costs and clarifies user and system requirements. An expert
panel evaluation highlighted the system's potential, particularly praising the
evaluation approach and the accessibility of the webbased simulation. This
paper outlines the infrastructure's architecture and implementation, providing
a foundation for future developments in social robotics evaluation and design.