To improve localization and pose precision of visual–inertial simultaneous localization and mapping (viSLAM) in complex scenarios, it is necessary to tune the weights of the visual and inertial inputs during sensor fusion. To this end, we propose a resilient viSLAM algorithm based on covariance tuning. During back-end optimization of the viSLAM process, the unit-weight root-mean-square error (RMSE) of the visual reprojection and IMU preintegration in each optimization is computed to construct a covariance tuning function, producing a new covariance matrix. This is used to perform another round of nonlinear optimization, effectively improving pose and localization precision without closed-loop detection. In the validation experiment, our algorithm outperformed the OKVIS, R-VIO, and VINS-Mono open-source viSLAM frameworks in pose and localization precision on the EuRoc dataset, at all difficulty levels.
High-fidelity static map construction is essential for reliable autonomous navigation, yet dynamic environments introduce severe artifacts caused by moving objects (also referred to as dynamic artifacts) in accumulated maps. While geometry-based methods perform well on dense point clouds, their performance notably degrades on sparse 16-beam LiDAR due to the “Sparsity Trap”: dynamic objects are frequently missed by ray-based geometry, and purely geometric cues fail in radiometrically ambiguous scenarios. To address this, we propose RI-DVP, a physics–geometry dual-driven framework. Unlike conventional approaches, RI-DVP first performs a physics-inspired radiometric normalization that compensates for range attenuation and incidence-angle effects to establish a consistent signal baseline. Subsequently, a Dual-Residual Aggressive Removal (DRAR) module jointly exploits geometric residuals—bounded by a range-dependent spatial uncertainty envelope—and calibrated intensity residuals to detect geometrically indistinguishable objects. To balance recall and precision, a Hierarchical Static Reversion strategy (HSR) employs two-stage recovery to retrieve large-scale structures and correct fine-grained artifacts via topology-based adhesion reasoning. Experiments on SemanticKITTI and custom sparse datasets demonstrate that RI-DVP outperforms state-of-the-art geometric baselines, improving Dynamic Accuracy by over 36 percentage points in sparse scanning scenarios using a VLP-16 LiDAR sensor (Velodyne Acoustics, Inc., Morgan Hill, CA, USA) compared to baselines that fail under the sparsity trap while achieving real-time performance at approximately 15.3 Hz.