ABSTRACT
Visual inertial navigation system (VINS) algorithms are considered state of the art for localization in cases where a global positioning system (GPS) is not available. The initialization procedure of VINS algorithms play a vital role in the overall accuracy and robustness of these algorithms. The estimation procedures used in VINS initialization can be classified as joint procedures vs the disjoint ones [Campos et al., 2020]. The joint estimation procedures build a closed form algebraic system of equations, which is typically solved using least squares based methodologies. On the other hand, the disjoint estimation procedures typically assume accurate measurement of up-to scale camera trajectory using monocular vision, which is used to estimate the inertial parameters. Regardless of these methods, there are certain degenerate motions which a wheeled mobile robot undergoes and fail to initialize, or initializes with very large errors, which affect the overall accuracy and robustness. In the recent past, considering these challenges, researchers have fused monocular visual, inertial, and wheel encoder measurements to estimate the robot states and initialize the state estimation, robustly to degenerate motions. However, the wheel encoder measurements can be faulty due to wheel slippages. Hence, when a VINS is augmented with wheel odometry (WO) one should take care of the errors arising due to wheel slippages and compensate for them. Considering the aforementioned challenges, this chapter introduces a method for augmenting VINS using slip compensated WO.
