Research Of Service Robots Task Planning With Hierarchical Task Network In Intelligent Space

With the development of robot technology, service robot has entered people’s lives increasingly. It plays an important role in daily service, especially in helping aged and disabled people. Task planning marks the intelligence of robot to some extent and is important to robot planning and decision-making. It is difficult for service robot to provide autonomous task planning and executing because the environment that robot works in is complex and dynamic. In recent years, the development of intelligent space technology makes environment information acquisition easier. Classical hierarchical task network planning is used widely in the field of service robot task planning for its similar characteristics of the human mind. With the support of intelligent space technology, this thesis aims to study robot task planning and online re-planning method in complex home environment based on the interaction between human and robot. The main works of this thesis are shown as followed:Firstly, this thesis studies the decomposition of robot task in home environment. Through decomposition, subtasks and atoms with different ranks and functions are generated. The definitions of the service task which can be understood by robot or intelligent space are given based on the characters of subtasks and atoms. The robot task domain description file is built. In the robot task domain description file, the tasks’ definition, tasks ’pre-conditions and tasks’ effects are described using the domain description language of JSHOP2 planner.Secondly, the robot task planning framework which integrate planner with intelligent space and robot is designed. After building domain description, problem description can be made according to user request and environment information. Then through using HTN planning algorithm, action list which can be executed by robot will be generated. Meanwhile, in order to increase robot’s intelligence and user’s satisfaction, user preference reasoning based on JESS expert system is added to the system.Thirdly, on the basis of task planning realized by JSHOP2 planner, the planning system is expanded and improved further to cope with the planning problems caused by environment changing and task changing. When environment changing, a feedback control mechanism which is designed from executor to planning system is added to realize task re-planning. When service task changing, the tasks’ priority is set and the multi-tasks scheduling algorithm is designed to realize multi-tasks scheduling. Thus the robot’s response ability of scene changing and emergency are improved.Fourthly, simulation tests and real scenario test are conducted. In simulation tests, in order to integrate planner system with intelligent space and robot, this thesis build the environment information database, improve the original JSHOP2 planner and design interaction interface. In real scenario test, the service robot which is independent developed by Service Robot Lab of Shandong University is chosen as actuator to verify the feasibility of the proposed method.