Research On The Intelligent Robotic Autonomous Writing Method

The operations of robot are one of the most important components and research topics in the field of intelligent robotics.On the other hand,Chinese character writing is a very complicated and meticulous work,writing has a high requirement with flexibility,stability and coordination and other requirements of the robot.Robotic writing ability is a research area that focuses on how to design algorithms to drive robot to write complex Chinese characters.In the current field of robotic Chinese characters writing,the robot suffer from the limited size of predefined font databases,monotonous style,and the effect is not satisfactory.In addition,the writing ability is obtained from a predefined system or a follow-up system,the level of involvement of human engineers is too large,nor does it apply to heavy or large robotic arms.Therefore,this paper focuses on:(1)improved automatic decomposing of Chinese characters,and(2)learning from human gestures,and stroke autonomous generation.Therefore,the process of the robotic learning skills of the writing ability is similar to human beings',to achieve the ability to learn strokes and write individual Chinese characters.This method not only reduces the complexity of the robot control algorithm,but also allows the robot can self-study and training,continue to learn and optimize the writing effect of strokes as human beings.In order to improve the way robots imitate writing Chinese characters,robots have the ability to deal with unfamiliar Chinese characters similar to human,perform self-learning while detecting the unknown words and to achieve writing.Firstly,in this dissertation,the existing automatic decomposing of Chinese characters mechanism has been improved,and two stroke learning methods of stroke autonomous generation have been achieved:(1)the robot learns the writing ability of different strokes through the human gestures,and applies the method to write individual Chinese characters;(2)the stroke autonomous generation algorithm is created to achieve the method that robot learning strokes by itself and writing.In the first stroke learning method,the corner detection technology is used,the system first captures a single Chinese character and decomposes it into a set of strokes.The robot's stroke trajectory writing ability is obtained by following the gestures demonstrated by a human demonstrator.A stroke classification method is used to recognize the decomposed strokes as the robot's writing trajectories.In the second method,a developmental evolutionary learning framework enables the robot to learn to write fundamental Chinese strokes.The framework first considers the learning process of robotic writing as an evolutionary procedure from easy to difficult.Then,a developmental learning mechanism called "Lift-Constraint,Act and Saturate" from developmental robotics determines how the robot learns simple tasks and learns more difficult tasks by using the learning results from the easy tasks.Developmental constraints including adjustment altitudes,mutation rates,and stroke trajectory points,determine the learning complexity of robotic writing.The developmental algorithm divided the evolutionary procedure into three developmental learning stages.Under various developmental constraint setup in each stage,the robot applies an evolutionary algorithm to handle basic shapes of the Chinese strokes,and then own the good quality writing ability.In addition,applies human arm gestures to establish a font database of English letters and Chinese character elementary strokes,then uses the created database and human gestures to write English words.The experiments illustrate that,the robot can write five basic strokes by following human gestures.Seven common Chinese characters are selected for the experiments.The selected characters can be decomposed correctly.The final writing result shows the robot successfully writes characters in a Chinese calligraphic style.On the other hand,the proposed framework is supported by experimental evaluations,demonstrating that the proposed framework allows a calligraphic robot to gradually write five fundamental Chinese strokes with good quality.In addition,the results also reveal a developmental pattern that is similar to one existing in human development.In contrast to an evolutionary algorithm without the developmental mechanism,the proposed framework has a faster speed to achieve good writing quality.