面向类人机器人动作规划的参数最优控制技术研究

项目名称： 面向类人机器人动作规划的参数最优控制技术研究

项目编号： No.61502256

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 安鹏

作者单位： 宁波工程学院

项目金额： 21万元

中文摘要： 类人机器人的动作规划如倒地控制等是机器人设计与应用中需要研究和解决的重要问题。项目旨在建立面向机器人动作规划的参数最优化可重构计算体系结构，提高机器人动作规划的参数最优化效率。具体内容包括，首先，建立类人机器人动作规划的参数最优化定点模型，研究参数最优化的算法特征；其次，针对模型内的算法瓶颈进行分析和优化，使其能采用可重构计算方法进行加速；再次，建立参数最优化可重构计算体系结构，对模型中的复杂算法做进一步优化，设计粗粒度自动细胞机结构和细粒度可重构加速部件，以提高处理效率。最后，构建面向类人机器人动作规划的性能测试平台。利用Nao型类人机器人和基于FPGA硬件平台的动作规划可重构体系结构，研究定点模型的真实性能，克服直接软件仿真和现场实测的缺陷。本项目建立的计算体系结构将更好地应用于机器人动作规划实践中，并促进类人机器人动作规划相关理论的发展。

中文关键词： 参数最优控制；类人机器人；可重构计算；自动细胞机

英文摘要： Humanoid robot motion planning, such as lying on the ground, is an important issue in the application of robots needs to be studied and solved. The project aims to establish the parameters for the robot motion planning optimization reconfigurable computing architecture，in order to improve parameters of the robot motion planning optimization efficiency. Specific content includes, first of all, to establish the parameters of humanoid robot motion planning optimization point model to study the optimization algorithm parameters characteristic; Secondly, bottlenecks within the model algorithm analysis and optimization, so that it can employ reconfigurable computing architecture accelerated; once again to establish the parameters for the robot motion planning class optimization reconfigurable computing architecture in the model, for further optimization of complex algorithms, reconfigurable components designed to accelerate the corresponding coarse-grained and fine-grained to improve processing efficiency. Finally, the class-oriented robot motion planning performance testing platform is established..Use Nao humanoid robot motion planning and FPGA-based reconfigurable architecture research could point the authenticity of the model to overcome the deficiencies and field simulation software directly measured. The project will establish parameters for the robot motion planning class optimization reconfigurable computing architecture, and effectively improve the development of humanoid robot motion planning related theories.

英文关键词： Parametric Optimal Control;Humanoid Robot;Reconfigurable Computing;Cellular Automation