Semantic SLAM for Collaborative Cognitive Workspaces

摘要

In order to create a meaningful interface between robots and humans, we advance the notion that a ”collaborative cognitive workspace” should form the basis for a dynamic information exchange between robots and their operators. One way to achieve this is through mapping the environment at the semantic level. Instead of traditional SLAM methods, which do not interpret sensor information other than at the geometric level, we propose that semantic mapping can provide the means to collaboratively build a shared understanding of the environment. By doing this, members of a heterogeneous robot-operator team can share information about the task and environment more naturally, using appropriate levels of abstraction that mitigate the perceptual and cognitive differences between team members.