Crowdsourcing Social Media For Robot Learning A Review

Abstract Methods designed to render capability of robots to learn on their own is a heavily studied area. If robots are to become a constitutional part of our society they must own the power to learn without direct guidance from a devoted user. Robot possessor never enjoyed the obligation of teaching their robot everything they know. The ability of a robot to use various resources in its environment will help its learning capabilities to be self-guided as well as independent. We investigate the use of social media crowdsourcing to allow a bot to access the huge information volume in gathering resources that are available on Twitter page. In distinction from others the robot will gather or record the basic human actions that are performed physically and are then upload that video on the Twitter account. It will now ask for the descriptions of those performed actions. Parameters of those recorded actions are used as input into a MC-SVM Multi-class Support Vector Machine classification algorithm which will enable the capability of robot to recognize the similar action in future.