Generating 3D human motions from textual descriptions is an important research problem with broad applications in video games, virtual reality, and augmented reality. Recent methods align the textual description with human motion at the sequence level, neglecting the internal semantic structure of modalities. However, both motion descriptions and motion sequences can be naturally decomposed into smaller and semantically coherent segments, which can serve as atomic alignment units to achieve finer-grained correspondence. Motivated by this, we propose SegMo, a novel Segment-aligned text-conditioned human Motion generation framework to achieve fine-grained text–motion alignment. Our framework consists of three modules: (1) Text Segment Extraction, which decomposes complex textual descriptions into temporally ordered phrases, each representing a simple atomic action; (2) Motion Segment Extraction, which partitions complete motion sequences into corresponding motion segments; and (3) Fine-grained Text–Motion Alignment, which aligns text and motion segments with contrastive learning. Extensive experiments demonstrate that SegMo improves the strong baseline on two widely used datasets, achieving an improved TOP 1 score of 0.553 on the HumanML3D test set. Moreover, thanks to the learned shared embedding space for text and motion segments, SegMo can also be applied to retrieval-style tasks such as motion grounding and motion-to-text retrieval.

Motion retargeting seeks to faithfully replicate the spatio-temporal motion characteristics of a source character onto a target character with a different body shape. Apart from motion semantics preservation, ensuring geometric plausibility and maintaining temporal consistency are also crucial for effective motion retargeting. However, many existing methods prioritize either geometric plausibility or temporal consistency. Neglecting geometric plausibility results in interpenetration while neglecting temporal consistency leads to motion jitter. In this paper, we propose a novel sequence-to-sequence model for seamless Spatial-Temporal aware motion Retargeting (STaR), with penetration and consistency constraints. STaR consists of two modules: (1) a spatial module that incorporates dense shape representation and a novel limb penetration constraint to ensure geometric plausibility while preserving motion semantics, and (2) a temporal module that utilizes a temporal transformer and a novel temporal consistency constraint to predict the entire motion sequence at once while enforcing multi-level trajectory smoothness. The seamless combination of the two modules helps us achieve a good balance between the semantic, geometric, and temporal targets. Extensive experiments on the Mixamo and ScanRet datasets demonstrate that our method produces plausible and coherent motions while significantly reducing interpenetration rates compared with other approaches. Code page: https://github.com/XiaohangYang829/STaR.

Aiming at the trajectory planning problem considering kinematic and dynamic manipulability optimizations, this article proposes an acceleration level manipulability maximization (ALMM) scheme to solve the problem in acceleration level for the first time. The manipulability index with nonlinear characteristics is reconstructed in acceleration level by a novel multilevel simultaneous processing strategy. The minimum joint velocity index is introduced to ensure the convergence of the system and maintain the velocity at a low level. Subsequently, the ALMM scheme, which includes pose maintenance, manipulability optimization, joint velocity minimization, and joint physical limit avoidance, is constructed and further formulated as a unified quadratic program. Then, a recurrent neural network with rigorously provable convergence is designed for the ALMM method. Simulations and physical experiments illustrate that the ALMM scheme can accomplish the manipulability optimization task excellently. Comparisons further verify the superiority of this scheme.

A kinematics control scheme in acceleration level is proposed to address the trajectory planning problem of redundant space manipulators under unknown loads or external forces. To achieve optimization of the joint torque from multiple sources, which is nonconvex relative to joint angles, quadratic programming (QP) is reconstructed through a novel multilevel simultaneous minimization scheme. This scheme, together with joint velocity norm minimization and physical limits avoidance formulated as the objective function and bound constraint into the QP problem. Subsequently, an extended recurrent neural network and intelligent numerical method, which are proven to converge, are established to solve the problem. Two trajectory tracking path simulations and experiments demonstrate the superb performance of this scheme when the manipulator is subjected to external forces or operating unknown loads. Comparisons with other schemes show that the proposed scheme is safer and more applicable.

The over-the-air (OTA) testing based on multi-probe anechoic chamber (MPAC) is an efficient solution to evaluate the performance of 5G multiple-input multiple-output (MIMO) capable devices, which can reconstruct the wireless channel within the lab in a controlled manner. For different test requirements, the probe layouts of the MPAC may be varied, bringing a lot of additional hardware overhead. In this paper, a novel design of the MPAC test system is proposed and constructed to meet various mainstream 5G OTA test solutions. By adopting switch matrixes and the three-dimensional (3D) quick probe selection algorithm, a 3D MPAC can be easily adjusted to different probe layouts. The solution can save at most 3/4 of the hardware port resources while ensuring the channel emulation accuracy. Channel validation and 5G terminals performance testing are carried out in the new system. The result comparison demonstrates the effectiveness of the probe simplification scheme and the system design.

Over-the-Air (OTA) testing, due to its capacity of reproducing the desired radio channels in lab environments, is considered as the promising testing solution for 5G multi-input-multi-output (MIMO) devices, especially for 5G massive MIMO base station (BS). A sectored multi-probe anechoic chamber (MPAC) OTA system for 5G massive MIMO BS testing has been proposed and discussed in many literatures. The probe selection is essential for 5G BS OTA testing, since the MPAC system cost is determined by the ports of channel emulator, i.e., the number of OTA probes. However, the classical convex-optimization-based probe selection method becomes complexity-prohibitive compared with the 2D MPAC system, due to the large amount of candidate probes. This paper proposes a novel probe selection for 5G BS OTA testing with MPAC setup based on the compressed sensing theory. The simulation results indicate that the proposed method can achieve high performance over single-cluster and multi-cluster channel models, in terms of high emulation accuracy and low computational complexity.

Testing solutions of new radio (NR) multiple-input-multiple-output (MIMIO) over-the-air (OTA) are discussed currently in the 3rd generation partnership project (3GPP), and the 16 evenly spaced probes layout has been determined for frequency range1 (FR1) MIMO OTA testing, which doubles the cost compared to the long term evolution (LTE) testing solutions. A novel probe selection algorithm based on standard FR1 channel model characteristic is proposed in this paper, which can reduce cost by at least 1/8 in measurement system. Simulation results show that good constructive accuracy can be achieved for target reference channel models. Channel models verification with selected probes is further performed including the power delay profiler (PDP), the temporal correlation function (TCF), and the spatial correlation function (SCF) parameters.

Optical flow estimation is an essential step for many real-world computer vision tasks. Existing deep networks have achieved satisfactory results by mostly employing a pyramidal coarse-to-fine paradigm, where a key process is to adopt warped target feature based on previous flow prediction to correlate with source feature for building 3D matching cost volume. However, the warping operation can lead to troublesome ghosting problem that results in ambiguity. Moreover, occluded areas are treated equally with non occluded regions in most existing works, which may cause performance degradation. To deal with these challenges, we propose a lightweight yet efficient optical flow network, named OAS-Net (occlusion aware sampling network) for accurate optical flow. First, a new sampling based correlation layer is employed without noisy warping operation. Second, a novel occlusion aware module is presented to make raw cost volume conscious of occluded regions. Third, a shared flow and occlusion awareness decoder is adopted for structure compactness. Experiments on Sintel and KITTI datasets demonstrate the effectiveness of proposed approaches.

Learning reliable motion representation between consecutive frames, such as optical flow, has proven to have great promotion to video understanding. However, the TV-L1 method, an effective optical flow solver, is time-consuming and expensive in storage for caching the extracted optical flow. To fill the gap, we propose UF-TSN, a novel end-to-end action recognition approach enhanced with an embedded lightweight unsupervised optical flow estimator. UF-TSN estimates motion cues from adjacent frames in a coarse-to-fine manner and focuses on small displacement for each level by extracting pyramid of feature and warping one to the other according to the estimated flow of the last level. Due to the lack of labeled motion for action datasets, we constrain the flow prediction with multi-scale photometric consistency and edge-aware smoothness. Compared with state-of-the-art unsupervised motion representation learning methods, our model achieves better accuracy while maintaining efficiency, which is competitive with some supervised or more complicated approaches.