In this paper, we aim to show strategies for improving graph interpretation skills at middle and high school students using OZOBOT® BIT, a small and relatively low-cost programmable robot which had been used to teach programming to young children. OZOBOT's speed can be controlled by drawing lines with colour codes, as well as through a visual block programming system. We show how to take advantage of these features of the robot by recording appropriate motions that students can explore by video analysis, manipulating data, characterizing the motion and associating graphs to predictions of physical models.