Intelligent robotic system for autonomous exploration and active SLAM in unknown environments

• Published in: 2017 IEEE/SICE International Symposium on System Integration (SII) pp. 651 - 656
• Main Authors: Zehui Meng, Hao Sun, Hailong Qin, Ziyue Chen, Cihang Zhou, Ang, Marcelo H.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.12.2017
• Subjects: Navigation; Planning; Simultaneous localization and mapping; Solid modeling
• DOI: 10.1109/SII.2017.8279295

In this paper, we present an intelligent robot system which is capable of autonomous exploration, navigation, and 3D mapping tasks in unknown environments. The system comprises of holonomic unmanned ground vehicle (UGV) and/or unmanned aerial vehicle agents implementing our novel optimized view planning paradigm that requires no prior environmental information of the environments. During the exploration, the robot agent conducts active SLAM to incrementally build a volumetric model (3-D mapping) of the environment using on-board sensors. The volumetric map is immediately employed for instantaneous navigation of the robot through our OctoMap-based navigation infrastructure, with valid motion plans generated by our sampling-based motion planner BIT*-H (integrated in the navigation interface) for cluttered environments. We test our system with simulation and experimental scenarios without giving prior knowledge of the environments. The results show that our system is capable of fast and complete autonomous exploration and mapping of the unknown environments.