Informative Mobile Robot Exploration for Radiation Source Localization with a Particle Filter

In this work, we consider the localization problem of an unknown radiation source with measurement uncertainty by using robotic systems in a geometric environment. We proposed the scheme for localization of a radioactive source using the particle filter with information gain-based exploration. The traditional method to localize the radiation is to use the gradient descent algorithm. However, such the algorithm may fail to work in the case of uncertain measurements, which lead to an inaccurate outcome. On the other hand, a standard particle filter can be used to deal with the measurement uncertainty, but the estimated intensity result may be unstable since it only uses the current measurement update as a likelihood function. To solve the problem of measurement uncertainty and unstable intensity result, we proposed an exploration method using the information gain with particle filter. The algorithm takes the information of the particles in the filter to estimate the possible actions for the robot. The expected information gain from those actions can be used to select the best possible action for the robot. The proposed method has been verified by the simulations. The proposed strategy can decrease the time it takes to finish the task comparing to the conventional methods such as the lawn mowing algorithm and source estimation seeking algorithm.