1、纯视觉SLAM开源方案:
稀疏地图
ORB SLAM v2 (单目、双目、RGB-D)
半稠密地图
LSD SLAM (单目、双目、RGB-D)
DSO (单目)
SVO(单目, 仅VO)
稠密地图
RGB-D SLAM V2 (RGB-D)
Kintinuous (RGB-D)
Elastic Fusion (RGB-D)
Bundle Fusion (RGB-D)
InfiniTAM (RGB-D)
RTAB-Map (RGB-D,双目,LIDAR)
2、多传感器融合 VINS (单目+IMU、双目+IMU)
OKVIS (单目+IMU、双目+IMU)
ROVIO (单目+IMU)
RKSLAM (单目+IMU)
Cartographer (LIDAR + IMU)
V-LOAM (单目+LIDAR)
3、和深度学习结合
CNN-SLAM: 将LSD-SLAM里的深度估计和图像匹配都替换成基于CNN的方法,并可以融合语义
VINet : Visual-inertial odometry as a sequence-to-sequence learning problem: 利用CNN和RNN构建了一个VIO,即输入image和IMU信息,直接输出估计的pose
3DMV: Joint 3D-Multi-View Prediction for 3D Semantic Scene Segmentation: 联合3D多视图预测网络在室内环境中进行RGB-D扫描的3D语义场景分割
ScanComplete: Large-Scale Scene Completion and Semantic Segmentation for 3D Scans: 将场景的不完整3D扫描作为输入,能够预测出完整的3D模型以及每个体素的语义标签
DeepVO: A Deep Learning approach for Monocular Visual Odometry Lightweight Unsupervised Deep Loop Closure: 用CNN解决闭环问题