GCNv2: Efficient Correspondence Prediction for Real-Time SLAM
This work addresses the need for efficient correspondence prediction in SLAM systems for robotics, such as drone control, but is incremental as it builds on the previous GCN method.
The paper tackles the problem of real-time SLAM by proposing GCNv2, a deep learning network for keypoint and descriptor generation that replaces ORB features, enabling efficient operation on embedded hardware like Jetson TX2 with comparable accuracy to its predecessor GCN.
In this paper, we present a deep learning-based network, GCNv2, for generation of keypoints and descriptors. GCNv2 is built on our previous method, GCN, a network trained for 3D projective geometry. GCNv2 is designed with a binary descriptor vector as the ORB feature so that it can easily replace ORB in systems such as ORB-SLAM2. GCNv2 significantly improves the computational efficiency over GCN that was only able to run on desktop hardware. We show how a modified version of ORB-SLAM2 using GCNv2 features runs on a Jetson TX2, an embedded low-power platform. Experimental results show that GCNv2 retains comparable accuracy as GCN and that it is robust enough to use for control of a flying drone.