在书写任务中的基于轨迹匹配的模仿学习

针对书写任务中运动轨迹较复杂的问题,引入基于轨迹匹配的模仿学习算法对书写轨迹进行表征和泛化,进而实现机器人书写技能的获取。机器人从示教者处获取示教数据,利用高斯混合模型( Gaussian mixture model, GMM)进行编码,学习示教行为的本质特征,通过高斯混合回归进行泛化处理,实现行为再现。实验结果表明：该方法具有良好的行为编码能力和抗干扰性,能够实现轨迹可连续的汉字书写,通过对GMM的扩展能够进行多任务学习,进而实现轨迹不可连续汉字的书写,泛化效果较好。

Imitation Learning Based on Trajectory Matching in the Writing Task

Aiming at the complexity of motion trajectory in the writing task, imitation learning based on trajectory matching was introduced to represent and generalize writing trajectory for the obtainment of writing skill. The robot acquires training data from a teacher and codes by Gaussian mixture model (GMM). Then, the essential feature of teaching behavior was learned and the motion trajectory was reconstructed by means of generalized output through Gaussian mixture regression (GMR). The results of simulation experiments show that this method possesses favourable ability of behavior coding and anti-interference performance. It can be used in the writing task of Chinese characters whose trajectory is continuous. Furthermore, the learning of Chinese characters’ writing skill with intermittent trajectory is realizable by multi-task learning extended from GMM and the generalization performance turns out to be good.