Meta-rooms: Building and maintaining long term spatial models in a dynamic world

Abstract

Abstract-We present a novel method for re-creating the static structure of cluttered office environments -which we define as the "meta-room" -from multiple observations collected by an autonomous robot equipped with an RGB-D depth camera over extended periods of time. Our method works directly with point clusters by identifying what has changed from one observation to the next, removing the dynamic elements and at the same time adding previously occluded objects to reconstruct the underlying static structure as accurately as possible. The process of constructing the meta-rooms is iterative and it is designed to incorporate new data as it becomes available, as well as to be robust to environment changes. The latest estimate of the meta-room is used to differentiate and extract clusters of dynamic objects from observations. In addition, we present a method for re-identifying the extracted dynamic objects across observations thus mapping their spatial behaviour over extended periods of time.

Autonomous Learning of Object Models on a Mobile Robot

“…For this we use the MetaRoom method [13], which performs segmentation based on the movement of clusters between observations. This method makes no prior assumptions such as CAD models or the physical features of objects.…”

“…Note that all steps in our process, up until the robot starts executing its viewplan, occur with the robot at the waypoint where it performs the most recent observation to update the MetaRoom. For completeness we briefly describe the Meta-Room method [13] here. First, we define the difference D between two point clouds P , Q as…”

On the use of unmanned aerial vehicles for autonomous object modeling

“…In [2], [3] we have shown how objects can be segmented out from a static background. We consider this output to be the input for our system.…”

“…We make use of a motion capture system for positioning the drone. In the past we have shown that a mobile robot can robustly segment out objects of interest in the environment through change detection [3]. This is what we envision as the means with which the objects to be observed are generated.…”

Life-long spatio-temporal exploration of dynamic environments

“…However, some dynamic objects do not move at the time of the mapping session and so the robot needs long-term observations to identify them. [16] propose to process several 3d point clouds of the same environment obtained over a period of several weeks to separate movable objects and refine the model of static environment structure at the same time.…”