Place Recognition with Event-based Cameras and a Neural Implementation\n of SeqSLAM

Abstract

Event-based cameras offer much potential to the fields of robotics and\ncomputer vision, in part due to their large dynamic range and extremely high\n"frame rates". These attributes make them, at least in theory, particularly\nsuitable for enabling tasks like navigation and mapping on high speed robotic\nplatforms under challenging lighting conditions, a task which has been\nparticularly challenging for traditional algorithms and camera sensors. Before\nthese tasks become feasible however, progress must be made towards adapting and\ninnovating current RGB-camera-based algorithms to work with event-based\ncameras. In this paper we present ongoing research investigating two distinct\napproaches to incorporating event-based cameras for robotic navigation: the\ninvestigation of suitable place recognition / loop closure techniques, and the\ndevelopment of efficient neural implementations of place recognition techniques\nthat enable the possibility of place recognition using event-based cameras at\nvery high frame rates using neuromorphic computing hardware.\n