Mobile Robot based Odor Path Estimation via Dynamic Window Approach

This paper addresses the problem of estimating the odor path which is most likely taken by the odor patch detected by the concentration sensor on a mobile robot moving in an indoor dynamic airflow environment. The odor path estimation is useful for plume tracing and odor source declaration. A novel algorithm for odor path likelihood mapping in the dynamic airflow environment is proposed. The algorithm has a low computation cost by importing the idea of dynamic window approach. Experiments are carried out on the mobile robot in which odor concentration sensor, airflow sensor, encoder and compass are equipped. To extract useable concentration information from the odor sensor, a practicable data preprocessing method is put forward. The experiment results in the indoor dynamic airflow environment show that the odor path can be well estimated online.