IDOL: A Framework for IMU-DVS Odometry using Lines

Abstract

In this paper, we introduce IDOL, an optimization-based framework for IMU-DVS Odometry using Lines. Event cameras, also called Dynamic Vision Sensors (DVSs), generate highly asynchronous streams of events triggered upon illumination changes for each individual pixel. This novel paradigm presents advantages in low illumination conditions and high-speed motions. Nonetheless, this unconventional sensing modality brings new challenges to perform scene reconstruction or motion estimation. The proposed method offers to leverage a continuous-time representation of the inertial readings to associate each event with timely accurate inertial data. The method's front-end extracts event clusters that belong to line segments in the environment whereas the back-end estimates the system's trajectory alongside the lines' 3D position by minimizing point-to-line distances between individual events and the lines' projection in the image space. A novel attraction/repulsion mechanism is presented to accurately estimate the lines' extremities, avoiding their explicit detection in the event data. The proposed method is benchmarked against a state-of-the-art frame-based visual-inertial odometry framework using public datasets. The results show that IDOL performs at the same order of magnitude on most datasets and even shows better orientation estimates. These findings can have a great impact on new algorithms for DVS.

Powerline Tracking with Event Cameras

“…Correctness Instance Persistence Hough transf. [9], [10], [11] + -+ Non parametric [12], [13] + --Spatio-temporal [14], [15] + + -Ours + + + and frequent changes in appearance in the stream of the events.…”

“…The difficulty of line tracking using events led previous works [9], [10], [11], [12], [13], [14], [15] to explore different techniques but pursuing a common goal: robustness. We identify the robustness of tracking a line as the combination of three desired characteristics.…”

“…When such assumption holds, lines describe a plane in the spatio-temporal event space. In [14], the normal vectors in the spatio-temporal space were computed for each incoming event using a local neighborhood. Events with similar normal vectors were clustered together to form lines.…”

Spatiotemporal Registration for Event-based Visual Odometry

“…A promising idea to realise asynchronous event-driven VO is to estimate a continuous-time (CT) camera trajectory from E. This implicitly allows the camera pose to be temporally interpolated and extrapolated, which in turn provides a basis to perform P1 and P2. Along the lines above, there have been efforts on event-driven VO [7], [8], [9], [10]. However, previous works either assume a known map [7], [8], restrict the camera to planar motions [9], or incorporate an IMU with preintegration [11] thereby achieving event-inertial VO [10].…”

“…Along the lines above, there have been efforts on event-driven VO [7], [8], [9], [10]. However, previous works either assume a known map [7], [8], restrict the camera to planar motions [9], or incorporate an IMU with preintegration [11] thereby achieving event-inertial VO [10].…”