Efficient Autonomous Robotic Exploration With Semantic Road Map in Indoor Environments

Abstract

This letter presents a novel and integrated framework for Next-Best-View (NBV) selection toward autonomous robotic exploration in indoor environments. A topological map, named semantic road map (SRM), is proposed to represent the explored environment during the exploration. The basic concept of the SRM is to construct a graph with nodes containing the exploration states and with edges satisfying the collision-free constraints. Especially, the SRM integrates both semantic and structure information of the environment, which possesses the beneficial properties of using a topological map in the exploration. It is worth noting that the proposed SRM is incrementally built along with the exploration process, thereby, avoiding the unnecessary reconsideration of the explored areas when constructing the topological map. Based on the SRM, a novel decision model with semantic information is presented for determining the NBV during the exploration. Moreover, the decision model takes into account both information gain and cost-to-go of a candidate NBV, which can be queried efficiently on the SRM, enabling the efficient exploration of the environment. The effectiveness and efficiency of the proposed system are assessed and demonstrated using both simulated and real-world indoor experiments.