A systematic mapping framework for backpack mobile mapping system in common monotonous environments. (30th June 2022)

• Record Type: Journal Article
• Authors: Bao, Sheng; Shi, Wenzhong; Chen, Pengxin; Xiang, Haodong; Yu, Yue
• Is Part Of: Measurement. Volume 197(2022)
• Publication Date: 2022-06-30
• Subjects: Mobile mapping system -- Monotonous environment -- PDR -- Factor graph -- Scene recognition -- LSTM; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2022.111243
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: A scene recognition method based on statistical features is proposed to detect monotonous environments. The concept of monotonous degree is proposed to measure how serious the monotonous environment is. A step detection method based on multiple long short-term memory neural networks is implemented. The accuracy of a backpack MMS supported by PDR is significantly improved in monotonous environments. Abstract: A mobile mapping system (MMS) is a widely-used platform to collect geospatial information. However, in the monotonous environment, current methods have inadequate performance in mapping accuracy because of lacking geometric features and constraints for point cloud alignment. We propose a systematic pedestrian dead reckoning (PDR) augmentation mapping framework for backpack MMS. The framework starts with data acquisition, followed by our proposed lightweight monotonous scene recognition method based on statistical features. An indicator is also proposed to measure monotonous degrees. Then, a step detection of PDR based on four-layer long short-term memory (LSTM) networks is implemented. Lastly, the PDR information is fused with the LiDAR odometry by a factor graph (FG). Experiments are conducted in two common monotonous environments, a tunnel and a long narrow alley. The results show that adding the PDR information can improve the mapping accuracy from meter-level to decimeter-lever or even centimeter-level in less serious monotonous conditions.