Self-Memory Prediction Approach for IBVS System under Uncertainty

摘要

This paper presents a self-memory prediction model to mitigate the effects of image based visual servoing (IBVS) system under uncertainty. The performance of IBVS system is easily influenced by different tasks, diverse environments and un-certain disturbances. Through building a self-memory prediction model to keep previous movement tendency in the every current movement, the framework of a self-memory is proposed for incremental self-learning capability. The main challenge in IBVS system is avoid abrupt change in the operation of the system. The proposed approach is processed from state pattern, and introduces the memory function which contain multiple historical observed values. According to past events, predict the current state and avoid abrupt changes in the every current movement. Simulations on IBVS system illustrate the effectiveness of the proposed method.