DF-SLAM: A Deep-Learning Enhanced Visual SLAM System based on Deep Local Features
This work addresses a key limitation in SLAM for driverless vehicles and intelligent robots, though it is incremental as it builds on existing SLAM frameworks with a deep learning enhancement.
The authors tackled the bottleneck of data association in traditional SLAM systems by replacing hand-made features with deep local feature descriptors from a neural network, resulting in improved efficiency and stability that outperforms popular systems in various scenes, including those with intense illumination changes, while maintaining real-time performance on GPU.
As the foundation of driverless vehicle and intelligent robots, Simultaneous Localization and Mapping(SLAM) has attracted much attention these days. However, non-geometric modules of traditional SLAM algorithms are limited by data association tasks and have become a bottleneck preventing the development of SLAM. To deal with such problems, many researchers seek to Deep Learning for help. But most of these studies are limited to virtual datasets or specific environments, and even sacrifice efficiency for accuracy. Thus, they are not practical enough. We propose DF-SLAM system that uses deep local feature descriptors obtained by the neural network as a substitute for traditional hand-made features. Experimental results demonstrate its improvements in efficiency and stability. DF-SLAM outperforms popular traditional SLAM systems in various scenes, including challenging scenes with intense illumination changes. Its versatility and mobility fit well into the need for exploring new environments. Since we adopt a shallow network to extract local descriptors and remain others the same as original SLAM systems, our DF-SLAM can still run in real-time on GPU.