Multi-dimensional pre-programmed and learned fine motor control for a humanoid robot

reinforcement learning;fine motor control;humanoid robot;iCub;proportional–integral–derivative (PID) control;balancing

Silver, Aaron ; Levinson ; Stephen E.

关键词: reinforcement learning;    fine motor control;    humanoid robot;    iCub;    proportional–integral–derivative (PID) control;    balancing;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/34413/Silver_Aaron.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

One of the biggest questions in modern electrical and computer engineering is that of computational intelligence.Can an artificially created mechanism, such as a humanoid robot, become intelligent, adaptable, and possess the ability to learn to classify the world in terms of language as humans do? This thesis discusses an effort to explore one small piece of this puzzle, the use of sensory-motor interaction with the physical world.We postulate that the ability to manipulate and interact with the environment around oneself isintegral to achieving intelligence and language.The first part of this study employs a sophisticated iCub humanoid robot and classical control techniques to determine if the robot is capable of performing the necessary fine-motor controlled tasks.The second part of this study goes on to determine if machine reinforcement learning techniques are able to give this robot the same capability.