The Battle for Filter Supremacy: A Comparative Study of the Multi-State Constraint Kalman Filter and the Sliding Window Filter

Abstract

Accurate and consistent egomotion estimation is a critical component of autonomous navigation. For this task, the combination of visual and inertial sensors is an inexpensive, compact, and complementary hardware suite that can be used on many types of vehicles. In this work, we compare two modern approaches to egomotion estimation: the Multi-State Constraint Kalman Filter (MSCKF) and the Sliding Window Filter (SWF). Both filters use an Inertial Measurement Unit (IMU) to estimate the motion of a vehicle and then correct this estimate with observations of salient features from a monocular camera. While the SWF estimates feature positions as part of the filter state itself, the MSCKF optimizes feature positions in a separate procedure without including them in the filter state. We present experimental characterizations and comparisons of the MSCKF and SWF on data from a moving hand-held sensor rig, as well as several traverses from the KITTI dataset. In particular, we compare the accuracy and consistency of the two filters, and analyze the effect of feature track length and feature density on the performance of each filter. In general, our results show the SWF to be more accurate and less sensitive to tuning parameters than the MSCKF. However, the MSCKF is computationally cheaper, has good consistency properties, and improves in accuracy as more features are tracked.

Monocular indoor localization techniques for smartphones

“…The motivation of Multi-State Constraint Filter (MSCKF) is the introduction of consecutive camera poses into state instead of observable feature landmarks, as it is first introduced by Nister [43], however this method does not incorporate inertial measurements. Sliding window-based solutions also appear in other papers [7].…”

A Review of Visual-Inertial Simultaneous Localization and Mapping from Filtering-Based and Optimization-Based Perspectives

“…Fusion Type Application OKVIS [43][44][45] optimization-based monocular tightly coupled SR-ISWF [46] filtering-based monocular tightly coupled mobile phone [47] optimization-based monocular tightly coupled [48] optimization-based Stereo tightly coupled MAV [49] optimization-based rgb-d loosely coupled Mobile devices [50] filtering-based monocular tightly coupled ROVIO [51] filtering-based monocular tightly coupled UAV [52] optimization-based monocular tightly coupled autonomous vehicle [53] filtering-based stereo tightly coupled [54] optimization-based stereo tightly coupled [55] optimization-based monocular tightly coupled [56] optimization-based stereo tightly coupled [57] filtering-based monocular loosely coupled robot [58] optimization-based rgb-d loosely coupled [59] filtering-based stereo loosely coupled VIORB [60] optimization-based monocular tightly coupled MAV [61] optimization-based rgb-d tightly coupled [62] filtering-based monocular loosely coupled AR/VR [63] filtering-based Multi-camera tightly coupled MAV [64] filtering-based monocular tightly coupled UAV VINS-mono [16][17][18] optimization-based monocular tightly coupled MAV, AR [65] optimization-based monocular tightly coupled AR [66] optimization-based monocular tightly coupled [67] filtering-based monocular tightly coupled MAV VINet [68] end-to-end monocular / deep-learning [69] optimization-based event camera tightly coupled S-MSCKF [26] filtering-based stereo tightly coupled MAV [70] optimization-based monocular tightly coupled MAV [71] optimization-based stereomonocular tightly coupled PIRVS [72] filtering-based st...…”

“…Thus they proposed modifications to the MSCKF algorithm, which ensure the correct observability properties without incurring additional computational costs. Clement [53] compared MSCKF and the sliding window filter (SWF). Its results showed the SWF to be more accurate and less sensitive to tuning parameters than the MSCKF.…”

Image-based camera localization: an overview

“…MSCKF algorithm is core algorithm of Google Project Tango https://get.google.com/tango/. [Clement et al 2015] compared two modern approaches: MSCKF and Sliding Window Filter (SWF). SWF is more accurate and less sensitive to tuning parameters than MSCKF.…”