发展型学习算法构建的移动机械手抓取能力的研究计算机技术专业论文.docxVIP

AbstractRobotic Abstract Robotic manipulator has a dexterous working ability in industrial and daily-life． The application range and adaptability in unstructured environments of static manipulators are limited．In contrast，mobile manipulators extend robot manipulators from a single workbench to much larger areas．Mobile robots with manipulators have significant usages in warehouses，logistics，military，and production lines．Therefore， the hand—eye coordination issue is worth of studying． The mapping between robotic visual space and hand space is nonlinear，and mobile platforms bring a new nonlinear topic，which is the main research field in mobile manipulator．The present research of robot hand—eye coordination focuses on how to establish the stimulus transformation from robotic visual space to hand space．Robotic hand-eye coordination is implemented by two different methods：firstly，the mathematical method，which employs inverse kinematics；secondly，the leaming approach，which uses mainly artificial neural networks．This paper introduces a reverse method：Build another channel that transfers stimulus signals from robotic hand space to visual space by using bi-directional stimulus delivery channels．Based on the forward and reverse channel，a human-like behavior pattem is imparted to the robot．In particular,when the robotic hand touches an unseen object，the reverse channel successfully drives the visual system to notice the unseen object． For the mobile manipulator hand-eye coordination in dynamic environments，the present research seldom considers robot’S autonomous feature in unstructured circumstance．This dissertation applies a developmental learning approach based on Lift—Constraint，Act and Saturate to mobile manipulator learning procedure．The robot learns novel abilities after releasing new constraint，which is set in sequence．The robotic hand-eye coordination ability is firstly developed in static environments，then III 万方数据 developed developed to dynamic environments．Two differ