Multi-modal object Re-Identification (ReID) aims to exploit complementary information from different modalities to retrieve specific objects. However, existing methods often rely on hard token filtering or simple fusion strategies, which can lead to the loss of discriminative cues and increased background interference. To address these challenges, we propose STMI, a novel multi-modal learning framework consisting of three key components: (1) Segmentation-Guided Feature Modulation (SFM) module leverages SAM-generated masks to enhance foreground representations and suppress background noise through learnable attention modulation; (2) Semantic Token Reallocation (STR) module employs learnable query tokens and an adaptive reallocation mechanism to extract compact and informative representations without discarding any tokens; (3) Cross-Modal Hypergraph Interaction (CHI) module constructs a unified hypergraph across modalities to capture high-order semantic relationships. Extensive experiments on public benchmarks (i.e., RGBNT201, RGBNT100, and MSVR310) demonstrate the effectiveness and robustness of our proposed STMI framework in multi-modal ReID scenarios.

This article presents an adaptive integral backstepping sliding mode controller (AIBSC) for a class of microgravity electromagnetic vibration isolation systems (MEVIS) that exhibit nonlinearity and uncertainty. To address the errors caused by the backstepping strategy, a finite-time command filter is applied. Additionally, to overcome acceleration disturbances, a fixed-time observer is used for estimation. Lyapunov theory is employed to prove the global asymptotic stability of the proposed controllers. Finally, simulations demonstrate that this controller ensures system stability while improving the isolation performance.

A fixed-time sliding mode controller is developed for the three-degree-of-freedom air-floating robot (AFR) in the presence of model uncertainties and external disturbances. By dynamics transformation, the complex nonlinear AFR dynamics are transformed into a second-order feedback decoupling model. A neural network enhanced by actor–critic learning structure is designed to handle model uncertainties, and the nonsingular fast terminal sliding mode controller can ensure all signals in the closed-loop system converge to a residual set around the origin within a fixed time. The system stability is proved by Lyapunov theory and the simulation results show the effectiveness of the proposed scheme.