Robot Interaction has always been a challenge in collaborative robotics. In tasks
comprising Inter-Robot Interaction, robot detection is very often needed. We
explore humanoid robots detection because, humanoid robots can be useful in many
scenarios, and everything from helping elderly people live in their own homes to
responding to disasters. Cameras are chosen because they are reach and cheap
sensors, and there are lots of mature two-dimensional (2D) and 3D computer
vision libraries which facilitate Image analysis. To tackle humanoid robot
detection effectively, we collected a data set of various humanoid robots with
different sizes in different environments. Afterward, we tested the well-known
cascade classifier in combination with several image descriptors like Histograms
of Oriented Gradients (HOG), Local Binary Patterns (LBP), etc. on this data set.
Among the feature sets, Haar-like has the highest accuracy, LBP the highest
recall, and HOG the highest precision. Considering Inter-Robot Interaction, it
is evident that false positives are less troublesome than false negatives, thus
LBP is more useful than the others.