Abstract:How to make simultaneous localization and mapping (SLAM) system to be robust and reliable in complex application scenarios has been widely regarded as a key part of automatic and practical robots. Robustness of SLAM system is one of the research focuses in SLAM field. To tackle this issue, we present multi-scene SLAM technique in this paper, which is a novel extensible SLAM framework by combing different SLAM systems facilitated by a scene detection method. Firstly, we propose a robust and extensible SLAM framework called SceneSLAM to enhance the self-adaptive performance of current SLAM systems. By model structure design, it is easy to utilize open-source SLAM systems to be SLAM models of SceneSLAM and modify the scene detection models to support complex application scenarios. Based on the scene detection results, it can schedule the SLAM models automatically. Secondly, aiming at indoor, outdoor and dark scene detection problems, we present a scene detection model based on convolutional neural network and Bias filter optimization. Experimental results show that the proposed scene detection model has reliable accuracy and stability in detecting indoor, outdoor and dark scenes. What’s more, considering sensor failure in indoor, outdoor and dark scenes, we design dynamic adaptive SLAM models to handle the issue. Map fusion is performed to achieve globally consistent localization and mapping when SLAM models scheduled. At last but not least, we build up a prototype system based on SceneSLAM to enhance the reliability of existing SLAM systems when the scene changes between indoor, outdoor and dark scenes. This system runs on a TurtleBot robot equipped with a Kinect sensor. The experimental results show that multi-scene SLAM technique proposed in this paper provides an effective solution to enhance the robustness of existing SLAM systems in dynamic environments.

Centralized Multi-Robot Visual SLAM and Scene Reconstruction

Multi-robot Collaborative SLAM and Scene Reconstruction Based on RGB-D Camera

Stereo Vision Based SLAM Using Rao-Blackwellised Particle Filter

Stereo Vision Based SLAM in Outdoor Environments

A Vision Based Multi-robot Cooperative Semantic SLAM Algorithm

Multi-Robot Collaborative Mapping with Integrated Point-Line Features for Visual SLAM

CCMD-SLAM: Communication-Efficient Centralized Multi-Robot Dense SLAM with Real-Time Point Cloud Maintenance

CCMD-SLAM: Communication-Efficient Centralized Multirobot Dense SLAM With Real-Time Point Cloud Maintenance

A Mobile Robot Visual SLAM System With Enhanced Semantics Segmentation

Distributed Monocular Multi-Robot Slam

CMDS-SLAM: Real-Time Efficient Centralized Multi-Robot Dense Surfel SLAM

Centralized Multi-Robot Collaborative LiDAR SLAM Utilizing Loop Closure Selection

CMDS-SLAM: Real-Time Eﬃcient Centralized Multi-Robot Dense Surfel SLAM

A Lightweight, Centralized, Collaborative, Truncated Signed Distance Function-Based Dense Simultaneous Localization and Mapping System for Multiple Mobile Vehicles

Edge Robotics: Edge-Computing-Accelerated Multirobot Simultaneous Localization and Mapping

Robust Perception-based Visual Simultaneous Localization and Tracking in Dynamic Environments

Edge Robotics: Edge-Computing-Accelerated Multi-Robot Simultaneous Localization and Mapping

Distributed and Collaborative Monocular Simultaneous Localization and Mapping for Multi-Robot Systems in Large-Scale Environments

Multi-robot collaborative dense scene reconstruction

SlideSLAM: Sparse, Lightweight, Decentralized Metric-Semantic SLAM for Multi-Robot Navigation

Dynamic Adaptive Simultaneous Localization and Mapping Technique for Scene Change