arXiv subjects

Olivier Brüls

Publications and source records attributed to Olivier Brüls.

6 recordsLinked to original sources

arXiv2026

Invariant Kalman filtering for extended pose estimation in multi-IMU articulated rigid-body systems

Accurate extended pose estimation (orientation, velocity, and position) for IMU-instrumented articulated rigid-body systems is a key challenge in robotics and human motion analysis. The invariant extended Kalman filter (IEKF) addresses this problem for a single rigid body with convergence guarantees and consistency under unobservability, but extending these properties to articulated systems is nontrivial: inter-body pose coupling prevents a direct application, and incorporating joint kinematic constraints within the invariant framework remains an open problem. To address this gap, we introduce the relative L-extended pose, a Lie group representation for kinematic-tree systems based on relative poses between bodies. With one IMU per body, it yields group-affine dynamics and allows joint constraints to be expressed in invariant form. We incorporate these constraints as noise-free pseudo-measurements within an iterated IEKF (IterIEKF), thereby preserving the convergence and consistency guarantees of invariant filtering. The absolute pose of each body is then recovered by chaining the relative blocks along the kinematic tree. Validated on both a UR5e robot and a human leg, the proposed IterIEKF outperforms all EKF, IterEKF, and absolute-pose IterIEKF baselines. It converges faster, exhibits lower run-to-run variability, and consistently achieves the lowest RMSE, with reductions of at least 50% compared to the second-best filter across all scenarios considered in this work.

cs.RO↗

arXiv2025

Iterated Invariant Extended Kalman Filter (IterIEKF)

We study the mathematical properties of the Invariant Extended Kalman Filter (IEKF) when iterating on the measurement update step, following the principles of the well-known Iterated Extended Kalman Filter. This iterative variant of the IEKF (IterIEKF) systematically improves its accuracy through Gauss-Newton-based relinearization, and exhibits additional theoretical properties, particularly in the low-noise regime, that resemble those of the linear Kalman filter. We apply the proposed approach to the problem of estimating the extended pose of a crane payload using an inertial measurement unit. Our results suggest that the IterIEKF significantly outperforms the IEKF when measurements are highly accurate.

eess.SY↗

arXiv2024

Invariant Kalman Filtering with Noise-Free Pseudo-Measurements

In this paper, we focus on developing an Invariant Extended Kalman Filter (IEKF) for extended pose estimation for a noisy system with state equality constraints. We treat those constraints as noise-free pseudo-measurements. To this aim, we provide a formula for the Kalman gain in the limit of noise-free measurements and rank-deficient covariance matrix. We relate the constraints to group-theoretic properties and study the behavior of the IEKF in the presence of such noise-free measurements. We illustrate this perspective on the estimation of the motion of the load of an overhead crane, when a wireless inertial measurement unit is mounted on the hook.

eess.SY↗

arXiv2023

Implicit representation priors meet Riemannian geometry for Bayesian robotic grasping

Robotic grasping in highly noisy environments presents complex challenges, especially with limited prior knowledge about the scene. In particular, identifying good grasping poses with Bayesian inference becomes difficult due to two reasons: i) generating data from uninformative priors proves to be inefficient, and ii) the posterior often entails a complex distribution defined on a Riemannian manifold. In this study, we explore the use of implicit representations to construct scene-dependent priors, thereby enabling the application of efficient simulation-based Bayesian inference algorithms for determining successful grasp poses in unstructured environments. Results from both simulation and physical benchmarks showcase the high success rate and promising potential of this approach.

cs.RO↗

arXiv2023

Simulation-based Bayesian inference for robotic grasping

General robotic grippers are challenging to control because of their rich nonsmooth contact dynamics and the many sources of uncertainties due to the environment or sensor noise. In this work, we demonstrate how to compute 6-DoF grasp poses using simulation-based Bayesian inference through the full stochastic forward simulation of the robot in its environment while robustly accounting for many of the uncertainties in the system. A Riemannian manifold optimization procedure preserving the nonlinearity of the rotation space is used to compute the maximum a posteriori grasp pose. Simulation and physical benchmarks show the promising high success rate of the approach.

cs.RO↗

arXiv2021

Simulation-based Bayesian inference for multi-fingered robotic grasping

Multi-fingered robotic grasping is an undeniable stepping stone to universal picking and dexterous manipulation. Yet, multi-fingered grippers remain challenging to control because of their rich nonsmooth contact dynamics or because of sensor noise. In this work, we aim to plan hand configurations by performing Bayesian posterior inference through the full stochastic forward simulation of the robot in its environment, hence robustly accounting for many of the uncertainties in the system. While previous methods either relied on simplified surrogates of the likelihood function or attempted to learn to directly predict maximum likelihood estimates, we bring a novel simulation-based approach for full Bayesian inference based on a deep neural network surrogate of the likelihood-to-evidence ratio. Hand configurations are found by directly optimizing through the resulting amortized and differentiable expression for the posterior. The geometry of the configuration space is accounted for by proposing a Riemannian manifold optimization procedure through the neural posterior. Simulation and physical benchmarks demonstrate the high success rate of the procedure.

cs.RO↗