6-DOF is kind of difficult.. You could do vision but that is hard to implement robustly. You say the UAVs are localizing themselves using GPS; are you using an IMU+Kalman filtering as well? For proximity detection you could try infrared/ultrasonic sensors, e.g. https://www.sparkfun.com/products/242. You could combine this with the Kalman filter to get a relative position estimate and get orientation from the Kalman filter directly.

Although I reckon if you are using GPS+IMU+Kalman then broadcasting the estimated state to each other (XBee?) would probably suffice, given that the GPS units are calibrated similarly.

6-DOF is kind of difficult.. You could do vision but that is hard to implement robustly. You say the UAVs are localizing themselves using GPS; are you using an IMU+Kalman filtering as well? For proximity detection you could try infrared/ultrasonic sensors, e.g. https://www.sparkfun.com/products/242. You could combine this with the Kalman filter to get a relative position estimate and get orientation from the Kalman filter directly.

Although I reckon if you are using GPS+IMU+Kalman then broadcasting the estimated state to each other (XBee?) would probably suffice, given that the GPS units are calibrated similarly. You could easily calculate the relative orientation by subtracting their orientation in the inertial frame.