基于姿态图及场景描述的服务机器人长期作业环境感知方法

项目名称： 基于姿态图及场景描述的服务机器人长期作业环境感知方法

项目编号： No.61473103

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 赵立军

作者单位： 哈尔滨工业大学

项目金额： 82万元

中文摘要： 近年来，服务机器人在助老助残领域需求迫切，由于环境长期感知问题制约着服务机器人智能作业，面向作业环境的长期地图构建问题成为当前移动机器人领域研究的热点。本项目针对现有技术存在数据量过大、抗动态性能差和实时更新等问题，开展基于姿态图与关键场景描述的长期移动作业环境感知的基础理论和方法研究。项目研究内容涉及长期环境动态性、地图长期构建可靠性、地图更新实时、计算复杂度和移动作业的适应性等关键科学问题，项目拟采用数据挖掘、聚类、状态图理论以及贝叶斯推断等方法对上述问题进行理论探索和实验验证。通过本项目的实施解决长期环境目标分类、状态图适应性约束及更新、闭环及关键场景识别、局部3D建模等多个核心环节，将在移动作业的长期定位及地图构建理论和方法上取得系列创新成果，并探索出适合于移动作业的室内自然环境低开销、实时性好、适应性强的描述方法，为面向服务机器人的室内长期应用奠定理论基础和方案支撑。

中文关键词： 智能机器人；信息融合；运动规划；环境感知

英文摘要： In recent years, with the ever-increasing demand for service robots in in the field of HelpAge and disabled，because the long-term perception of environmental issues limit the service robot mobile intelligence operations, mapping of long term operating environment has become a hot topic in the field of mobile robot research . Based on problems of related technologies such as data size too large, poor ability to dynamic objects and realtime updates, The project will study on the basic theory and methods to long-term mobile operation environment perception based on pose gragh and key scene discription. Research project involving long-term environmental dynamics, stability of long-term mapping, the map updating in real time, computational complexity, adaptability of mobile operations and other key scientific issues, the project intends to use data mining , clustering, pose graph theory and Bayesian inference , etc. theoretical approach to theoretical exploration and experimental validation. Performances of this project will solve the long-term environmental objects classification, adaptive constraint and update pose gragh, loop lose and key scenes recognition,local 3D modeling and other core parts, which can realize theoretical and methodological series of innovations for long-term localization and mapping. Finally，we propose a novel method with low cost, real-time, adaptable description to mobile operating natural environment for indoor service robot that will give the theoretical solid basis and programs support for robots long-term application.

英文关键词： intelligent robots;information confusion;motion planning;environment perception