Cascone et al. designed EventsFormer, a Transformer-based architecture that takes voxel-grid representations of event streams over short time windows and outputs a 3-DoF body-frame translation estimate and per-axis uncertainty.
Learned event-inertial odometry uses a Transformer to predict motion directly from event streams
The TLEIO system fuses learned translation estimates with IMU data in a filtering framework, achieving lower error and faster runtime than optimization-based event odometry.
Big Tech
Matteo Cascone · David Jensen · Alessandro Pirini · Angelo Stefanini · Pietro Venè · Marta Tintoré Gazulla · +1 more
ETH Zürich · Google
Research Digest··2 min read
The authors introduce Tight Learned Event-Inertial Odometry (TLEIO), a pipeline whose learned front-end, EventsFormer, predicts short-window relative translation and per-axis uncertainty from voxelized event streams in a single forward pass.
The Daily Edition
The front page in your inbox each weekday morning: the lead, the picks, and why each one earned its place.