Event-Based Visual Inertial Odometry Using Smart Features
Abstract
Event-based cameras are a novel type of visual sensor that operate under a unique paradigm, providing asynchronous data on the log-level changes in light intensity for individual pixels. This hardware-level approach to change detection allows these cameras to achieve ultra-wide dynamic range and high temporal resolution. Furthermore, the advent of convolution neural networks (CNNs) has led to state-of-the-art navigation solutions that now rival or even surpass human engineered algorithms. The advantages offered by event cameras and CNNs make them excellent tools for visual odometry. A visual odometry pipeline was implemented with a front-end network for generating event-frames that were fed into a multi-state constraint Kalman filter (MSCKF) back-end, which utilized features and descriptors generated by a CNN. This pipeline was tested on a public dataset and data collected from an ANT Center UAV flight test.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 26, 2020
- Accession Number
- AD1102920
Entities
People
- Zachary P Friedel
Organizations
- Air Force Institute of Technology