This paper describes a method for reliably generating traffic conflicts in human-in-the-loop (HITL) experiments. Research, development and pilot training on systems such as TCAS or ACAS-X require HITL simulation studies, which have been historically limited by their ability to create desired collision avoidance trajectories. With the present method, pilots are free to control their own trajectory within the limits of air traffic instructions. The simulator adapts to any variation in the ownship's trajectory as it generates intruder trajectories. The method not only results in an RA, but also provides the ability to reliably control the type of initial resolution advisory (climb, descend,...) and any modifications (reversal, strengthen). Moreover, the generated intruder trajectories remain realistic and fade in with surrounding traffic. Therefore, pilots' reactions can be accurately studied in realistic settings and with minimal variation in the experimental conditions. Results from a recent study are presented, illustrating success rates in achieving desired initial resolution advisory and any modifications to it.