机器人系统中的在线环境参数辨识

In order to investigate the techniques ofonline parameter estimation for robot systems, the modeling accuracies of different methods in environment modeling are compared bycross-validation in various experimental environment. Common estimation algorithms are reviewed and compared, discussions are given in terms of convergence and tracking ability. We first evaluate the changes of model accuracy during online estimation by cross-validation to give a more comprehensive comparison between different estimation algorithms' performance. Practical guidelines for applications of online environment parameter estimation techniques are provided.