Control of hyper-redundant robot using QDSEGA

Abstract

We consider a flexible autonomous system. To realize the system, we employ H?.per-rednndant system (It is flexible hardware systm) and Reinforcement learning controller "QDSEGA'; (It is a flexible software system). In this paper we apply QDSEGA to controlling of Hyper-redundant robot. To demonstrate the effectiveness, a task of acquisition of lmoniotion patterns is applied to a niulti-legged format.ioii and a snakelike formation, as a result effective locomotion hss been obtained.

Logical-Probabilistic Method for Controlling Modular Robots

“…в интеграции с другими известными методами машинного обучения (Reinforcement Learning, нейронные сети и др.) [7,8,[11][12][13]17,18]. Однако эволюционные методы имеют серьезные ограничения, связанные с необходимостью наличия популяции роботов, что не позволяет проводить обучение и адаптацию в режиме реальной работы [14].…”

System errors quantitative analysis of sample-scanning AFM

“…Many researches have been done on tube scanner errors caused by piezoceramics intrinsic characteristics such as nonlinearity and hysteresis, and various calibration and nonlinearity correction methods have also been proposed [6][7][8][9]. Some models mainly to investigate scanner dynamics have also been proposed, e.g., a complex dynamic model considering the coupling between motions in different axes has been presented in Ref.…”