In Robotics Education (RE), the hands-on experience with troubleshooting problems is seen as a good catalyst to enhance the participants' problem-solving skills. Based on the pedagogical technique of collaborative learning and pair programming, pair learning is an emerging and potential method in RE, which means that students collaborate in pairs to construct, build, and program a robot under scripted but switchable roles. As a special collaborative learning, can pair learning also improve students' troubleshooting performance in RE? Therefore, we conducted a comparison experiment (pair learning versus individual learning) in two classes at a Robot Summer Camp for high school students. The results indicated that (a) in the process of learning making robot artifacts, students in pair learning group (PLG) had a higher success rate in troubleshooting than that of individual learning group (ILG), but failed in other indicators and (b) in the summative test, the troubleshooting performance of students in PLG was similar with that of ILG. The findings showed that, in terms of troubleshooting, collaborative (pair) learning does not always has the superiority over individual learning in RE, which implicates other factors to be explored in future research. Implications for teaching are also discussed in this exploratory study.