Observation planning for map updating tasks by predicting changes in environments.

Kanji Tanaka,Hongbin Zha,Tsutomu Hasegawa

DOI: https://doi.org/10.1109/IROS.1999.813022

1999-01-01

Abstract:For robots working in a dynamic environment, there is a great need for a global environment map that should be updated timely. The paper presents an observation planning method for the map updating purpose. The task assigned to the robot is to detect changes of environment at planned observation positions and update the map into the newest version. After the observation at a current position, the robot gets quantitative estimations on the changes that may occur at possible sequences of future observations. Then, it evaluates the sequences for the purpose of map updating and selects the most favored next position as the planned one. Efforts are made in searching the optimal position by a dynamic programming algorithm with constraints from previous searching results.