This study investigates the use of Virtual Reality (VR) technology as a tool to assess and improve personal safety perceptions among vulnerable public transport users. High-fidelity VR environments were developed in Unreal Engine 5 using pre-built assets from the Unreal Marketplace to replicate a sample university bus stop corridor. Eight infrastructure interventions, derived from the SHE CAN tool, were embedded in the environment to test their influence on perceived safety. Participants, recruited through community partnerships, completed both VR simulations and physical walkthroughs of the same corridor. Data collection combined continuous verbalised threat ratings, eye-tracking and behavioural logging, and semi-structured interviews. A novel graphical output, the Safety Cloud, was generated to visualise georeferenced safety perceptions using an intuitive weather-map analogy. Preliminary findings suggest that VR-based safety ratings align closely with real-world perceptions, validating VR as a research tool in African transport contexts. Early results also indicate gender-based differences, with female participants reporting greater discomfort in poorly lit or visually obstructed areas, mirroring findings from earlier studies (Vanderschuren et al., 2017; Newlands et al., 2023). By integrating immersive simulations with traditional fieldwork, this study introduces VR as a cost-effective and scalable method for testing interventions before real-world investment. The Safety Cloud framework further enhances communication of safety perceptions for planners and policymakers, supporting more inclusive and equitable transport design in South Africa.