During the past few years, increasing robotic manipulator techniques have been employed in industrial applications and our daily lives. One of the most difficult problems in robotic research is how to achieve a human-like working manner in its workspace, such as the distinctive capability of skills learning and friendly desirable interaction control (Yang et al. in IEEE Trans Indus Inf 13(3):1162–1171, 2016 [1]; Liu et al. in Neurocomputing 275:73–82, 2018 [2]; Zhang et al. IEEE/ASME Trans Mechatr 25(1):95–107, 2019 [3]). Such abilities are of great importance to enable robots to respond more intelligently and be able to fulfil more complex, dexterous and versatile tasks in various fields such as industrial applications and service areas.