Reinforcement learning coupled with Neural Networks has been demonstrated to perform well in robotic manipulation tasks. Yet they require large volume of sample data which are trained using huge amount of computing resources. We propose an End-to-End Neural Network architecture based robotic system that can be deployed on embedded platforms such as a Raspberry Pi to perform robotic grasping. The proposed method potentially solves the exploration-exploitation dilemma even under undecidable scenarios.
P. Sreedhar and Palaniswamy, S., “Robotic Grasper based on an End-to-End Neural Architecture using Raspberry Pi”, in 24th annual International conference on Advanced Computing and Communications (ADCOM 2018), IIITB, Advanced Computing and Communications Society, IISc, Bangalore, 2018.