This paper presents an algorithm to improve state estimation for legged robots. Among existing model-based state estimation methods for legged robots, the contact-aided invariant extended Kalman filter defines the state on a Lie group to preserve invariance, thereby significantly accelerating convergence. It achieves more accurate state estimation by leveraging contact information as measurements for the update step. However, when the model exhibits strong nonlinearity, the estimation accuracy decreases. Such nonlinearities can cause initial errors to accumulate and lead to large drifts over time. To address this issue, we propose compensating for errors by augmenting the Kalman filter with an artificial neural network serving as a nonlinear function approximator. Furthermore, we design this neural network to respect the Lie group structure to ensure invariance, resulting in our proposed Invariant Neural-Augmented Kalman Filter (InNKF). The proposed algorithm offers improved state estimation performance by combining the strengths of model-based and learning-based approaches. Project webpage: https://seokju-lee.github.io/innkf_webpage

Recently, legged robots have been used for various purposes, such as exploring unknown terrain or interacting with the world. For control and planning legged systems during interactive operations, it is essential to estimate and respond to external forces. However, in legged system, it becomes difficult to estimate forces due to highly dynamic situations. There are several studies that use a force sensor on the foot and end effector, but these approaches have disadvantages in terms of cost and sustainability. Therefore, in this paper, we propose an improved method for estimating external forces without a force sensor. First, each leg force was obtained using the system dynamics of the robot with a disturbance observer. Then, by preintegration, it was tightly coupled with other sensors to estimate the pose and external force simultaneously. Despite the impact and slip, we estimate external forces accurately in standing and walking motions. Moreover, we compared pose estimation performance with VINS-Mono [1], and there is no significant accuracy degradation in spite of highly dynamic force residual.

We propose a novel snake algorithm for object segmentation that can overcome the limitations found in cluttered backgrounds, overlapped object placements, and the problems of initial snake points. Our newly designed algorithm is developed by using disparity information taken from several sets of stereo images. The experiment results have exhibited a better performance over the well-known snake algorithm in terms of segmentation accuracy. It is also shown that our proposed new segmentation algorithm can be extended to the tracking of an object visible boundary and occlusion detection in the 3D disparity space.

The basic ideas for realistic and real-time animation of 3D character are described through three steps. First, we just choose two deformations among all kinds of deformations that happen in the human body, because these deformations are globally noticed and can be performed in real-time. Second, we decide what body parts of a character are deformed from the anthropometrics data. Last, we propose simple and fast implementations for two deformations. Using proposed ideas, the authoring tool that make a deformable character from a rigid polygon model and test deformation of it, is introduced. And the animation player to animate a deformable character generated from the authoring tool is also introduced.