In order to relieve the user cognitive burden as well as time consumption of task execution for the “wheelchair mounted robotic arm” application, we proposed a "Task knowledge representation" method that is able to model a large number of tasks coming across in activity of daily livings. Based on the mathematically represented task knowledge, a “Dynamical exploring method” was applied to explore the feasible and optimal robot action sequences that are executed automatically to accomplish the task execution. An integration framework to manage the system complexity is also introduced based on the RT middleware technology. Experiment was conducted to validate our method through the execution of a robotic scenario.
展开▼