Deep learning for 2D scan matching and loop closure

Although 2D LiDAR based Simultaneous Localization and Mapping (SLAM) is a relatively mature topic nowadays, the loop closure problem remains challenging due to the lack of distinctive features in 2D LiDAR range scans. Existing research can be roughly divided into correlation based approaches e. g. scan-to-submap matching and feature based methods e. g. bag-of-words (BoW). In this paper, we solve loop closure detection and relative pose transformation using 2D LiDAR within an end-to-end Deep Learning framework. The algorithm is verified with simulation data and on an Unmanned Aerial Vehicle (UAV) flying in indoor environment. The loop detection ConvNet alone achieves an accuracy of 98. 2% in loop closure detection. With a verification step using the scan matching ConvNet, the false positive rate drops to around 0. 001%. The proposed approach processes 6000 pairs of raw LiDAR scans per second on a Nvidia GTX1080 GPU.

Authors

• Jiaxin Li
• Huangying Zhan
• Ben M. Chen
• Ian D. Reid 0001
• Gim Hee Lee

Keywords

• Feature extraction
• Laser radar
• Training
• Two dimensional displays
• Machine learning
• Simultaneous localization and mapping
• Deep Learning
• Loop Closure
• Scan Matching
• Convolutional Neural Network
• False Positive Rate
• Unmanned Aerial Vehicles
• Indoor Environments
• Deep Learning Framework
• LiDAR Scans
• Scan Pairs
• Loop Detection
• High Response
• Laser Scanning
• Convolutional Layers
• Computer Vision
• Feature Maps
• Long Short-term Memory
• Recurrent Neural Network
• Maximum Velocity
• Nearest Neighbor Search
• Iterative Closest Point
• Robot Operating System
• Term Frequency-inverse Document Frequency
• Receptive Field
• Cartesian Space
• Softmax Function
• Stack Of Convolutional Layers
• Alignment Errors
• Computer Vision Community