Event-Based Visual Inertial Odometry
Citations
697 citations
Cites background or methods from "Event-Based Visual Inertial Odometr..."
...ks are extracted from the events, and then these point trajectories on the image plane are fused with IMU measurements using state-of-the-art VIO algorithms, such as [207], [208], [209]. For example, [119] tracked features using [116], and combined them with IMU data by means of the Kalman filter in [207]. Recently, [49] proposed to synthesize motion-compensated event images [115] and then detect-and-tr...
[...]
... exploit the fine temporal information of individual events for estimation, and therefore tend to depart from traditional computer vision algorithms [23], [26], [32], [49], [115], [116], [117], [118], [119], [120]. The review [7] quantitatively compares accuracy and computational cost for frame-based versus event-driven optical flow. Events are processed differently depending on their representation. Som...
[...]
...ilt in [116] from motioncompensated events, producing point-set–based templates to which new events are registered. These features allowed to tackle the moving camera scenario in natural scenes [46], [119]. Also in this scenario, [49] proposed to apply traditional feature detectors [167] and trackers [168] on patches of motion-compensated event images [115]. Hence, motioncompensated events provide a us...
[...]
... [204]18, [182], [205], [250], [251]. The most popular one is described in [205], which has been used to benchmark visual odometry and visual-inertial odometry methods [26], [49], [50], [115], [118], [119], [203]. This dataset is also popular to evaluate corner detectors [122], [123] and feature trackers [46], [159]. Datasets for recognition are currently of limited size compared to traditional compute...
[...]
358 citations
Cites methods from "Event-Based Visual Inertial Odometr..."
...The extended Kalman filter backend in [29] implements this formulation of the MSCKF for event-based camera inputs, but has been adapted to feature tracks from standard cameras....
[...]
280 citations
Cites methods from "Event-Based Visual Inertial Odometr..."
...The authors in [17] and [18] proposed novel methods to perform feature tracking in the event space, which they extended in [19] and [20] to perform visual and visual inertial odometry, respectively....
[...]
277 citations
212 citations
References
11,283 citations
"Event-Based Visual Inertial Odometr..." refers background or methods in this paper
...In each EM step, the template point sets {l̃j} are subsampled using sphere decimation [8], with radius 1 pixel....
[...]
...In Table 1, we present the mean position error as a percentage of total distance traveled and rotation error over distance traveled for each sequence, which are common metrics for VIO applications [8]....
[...]
8,432 citations
"Event-Based Visual Inertial Odometr..." refers methods in this paper
...The actual corner detection is performed with FAST corners [17], with the image split into cells of fixed size, and the corner with the highest Shi-Tomasi score [18] within each cell being selected, as in [5]....
[...]
1,847 citations
"Event-Based Visual Inertial Odometr..." refers methods in this paper
...The actual corner detection is performed with FAST corners [17], with the image split into cells of fixed size, and the corner with the highest Shi-Tomasi score [18] within each cell being selected, as in [5]....
[...]
1,814 citations
"Event-Based Visual Inertial Odometr..." refers methods in this paper
...The actual corner detection is performed with FAST corners [17], with the image split into cells of fixed size, and the corner with the highest Shi-Tomasi score [18] within each cell being selected, as in [5]....
[...]
1,435 citations
"Event-Based Visual Inertial Odometr..." refers background or methods in this paper
...We then left multiply r by the left null space, A, of the feature Jacobian, HF , as in [14], to eliminate the feature position up to a first order approximation:...
[...]
...Similar to the MSCKF [14], we eliminate the depth from the measurement equations so that we do not have to keep triangulated features in the state vector....
[...]
...For compactness, we do not expand on the fine details of the filter, and instead refer interested readers to [13] and [14]....
[...]
...As in [14], we perform one final step to reduce the dimensionality of the above residual....
[...]
...6 then employs an Extended Kalman Filter with a structureless vision model, as first introduced in [14]....
[...]