The existence of emergent properties, desirable or undesirable, makes a system harder to analyze and design, and requires a formal approach for detecting and reasoning about its causes and nature. The research effort presented in this extended abstract focuses on exploring emergent behaviors in a multi-agent dynamical system with the intent of reduction in complexity of detection of such unexpected behaviors. Our approach relies on the theory of similitude (or physical similarity), where the main idea is that similar behaviors occur when the values of the system variables are in a specific relation. These relations, captured using dimensionless quantities, define a hypersurface in the space spanned over the system variables, which in turn can be used to measure the distance to potential undesirable behaviors. We use similitude theory to detect undesirable emergent behaviors in swarms of UAVs (Unmanned Aerial Vehicles), treated as a complex dynamical system. Since this is a part of ongoing research, future directions are also discussed.