Combined denoising method of acceleration signal during mobile robot collision

Abstract Collision safety of mobile robot is one the important indexes evaluation. The acceleration signal can completely feed-back state collision process, but it also vulnerable to noise interference such as vibration and impact, which difficult be used for direct analysis. In order solve problem mixing with impact periodic during collision, a combined filtering reduction method based on empirical mode decomposition (EMD), improved wavelet threshold dynamic sliding window proposed. And adaptive function constructed. coefficients are automatically adjusted using information degree level, adjustment strategy length designed. effectively extracted in process strong collision. experimental results show that this has better effect than other methods.