Semantic Linking Maps for Active Visual Object Search
This work addresses the challenge of active visual object search for mobile robots in common human environments, representing an incremental improvement by integrating probabilistic spatial reasoning into existing search methods.
The paper tackles the problem of enabling mobile robots to efficiently search for unseen target objects in human environments by leveraging probabilistic spatial relations between landmark and target objects, resulting in a proposed Semantic Linking Maps (SLiM) model and hybrid search strategy that demonstrates efficiency in simulated and real-world robot experiments.
We aim for mobile robots to function in a variety of common human environments. Such robots need to be able to reason about the locations of previously unseen target objects. Landmark objects can help this reasoning by narrowing down the search space significantly. More specifically, we can exploit background knowledge about common spatial relations between landmark and target objects. For example, seeing a table and knowing that cups can often be found on tables aids the discovery of a cup. Such correlations can be expressed as distributions over possible pairing relationships of objects. In this paper, we propose an active visual object search strategy method through our introduction of the Semantic Linking Maps (SLiM) model. SLiM simultaneously maintains the belief over a target object's location as well as landmark objects' locations, while accounting for probabilistic inter-object spatial relations. Based on SLiM, we describe a hybrid search strategy that selects the next best view pose for searching for the target object based on the maintained belief. We demonstrate the efficiency of our SLiM-based search strategy through comparative experiments in simulated environments. We further demonstrate the real-world applicability of SLiM-based search in scenarios with a Fetch mobile manipulation robot.