南京大学学报(自然科学版)
南京大學學報(自然科學版)
남경대학학보(자연과학판)
JOURNAL OF NANJING UNIVERSITY(NATURAL SCIENCES)
2010年
2期
131-139
,共9页
吉建民%陈小平%姜节汇%靳国强%王锋
吉建民%陳小平%薑節彙%靳國彊%王鋒
길건민%진소평%강절회%근국강%왕봉
自主机器人%人机协调%个性化%自然语言处理%answer%set逻辑程序设计
自主機器人%人機協調%箇性化%自然語言處理%answer%set邏輯程序設計
자주궤기인%인궤협조%개성화%자연어언처리%answer%set라집정서설계
autonomous robot%human-machine coordination%individualization%natural language processing%answer set programming
本文提出一种支持个性化协调的服务机器人体系结构(individualized coordinationarchitecture,ICA).主要动机是通过自然语言人机对话获取用户的个人特性和其他信息,通过对这些信息进行自动推理和规划,实现利用个人特性的自动问题求解,并满足家庭环境对服务机器人的应用要求.本文着重介绍ICA的主要部件的功能及其相互衔接方式,描述任务规划的机制和实现手段,并通过一个实例说明在一个初步实现的原型系统中从自然语言到任务规划的完整过程.
本文提齣一種支持箇性化協調的服務機器人體繫結構(individualized coordinationarchitecture,ICA).主要動機是通過自然語言人機對話穫取用戶的箇人特性和其他信息,通過對這些信息進行自動推理和規劃,實現利用箇人特性的自動問題求解,併滿足傢庭環境對服務機器人的應用要求.本文著重介紹ICA的主要部件的功能及其相互銜接方式,描述任務規劃的機製和實現手段,併通過一箇實例說明在一箇初步實現的原型繫統中從自然語言到任務規劃的完整過程.
본문제출일충지지개성화협조적복무궤기인체계결구(individualized coordinationarchitecture,ICA).주요동궤시통과자연어언인궤대화획취용호적개인특성화기타신식,통과대저사신식진행자동추리화규화,실현이용개인특성적자동문제구해,병만족가정배경대복무궤기인적응용요구.본문착중개소ICA적주요부건적공능급기상호함접방식,묘술임무규화적궤제화실현수단,병통과일개실례설명재일개초보실현적원형계통중종자연어언도임무규화적완정과정.
A service robot architecture supporting individualized coordination is proposed. In this approach, user features and other information will be acquired through human-robot dialogue in natural languages. The robot reasons about and makes plans with the information, so that automatic, individualized problem-solving can be carried out under real-world environments. We present the main modules and their interfaces of the architecture, the mechanism of task planning, and a case study which includes the entire working process from natural language processing(NLP) to task planning in an implemented prototype system, expected to work with humans in, say, ordinary home environments. Standard NLP techniques such as syntactic parsing have been implemented in the prototype system, together with semantic analyzing developed by the authors. With these NLP techniques, the user's commands and descriptions about the environment expressed in natural languages are transferred into logical forms in segmented discourse representation theory(SDRT) and finally in answer set programming(ASP), which are ready for the task planning, through a 5-steps procedure. In task planning module, ASP technique is employed, making it possible and feasible for the robot to conduct task planning and reasoning about the actions and changes in a unifying way. With some good coding, according to the authors' experiments with a real home robot, the task planning module can work for some typical tasks in real-time. This work provides a basis for investigations into further challenges in individualized coordination.