A longstanding question in computer vision concerns the representation of 3D shapes for recognition: should be represented with descriptors operating on their native formats, such as voxel grid or polygon mesh, can they effectively view-based descriptors? We address this context learning to recognize from a collection rendered views 2D images. first present standard CNN architecture trained shapes' independently each other, and show that shape recognized even single view at an accuracy far...

We present a network architecture for processing point clouds that directly operates on collection of points represented as sparse set samples in high-dimensional lattice. Naively applying convolutions this lattice scales poorly, both terms memory and computational cost, the size increases. Instead, our uses bilateral convolutional layers building blocks. These maintain efficiency by using indexing structures to apply only occupied parts lattice, allow flexible specifications structure...

We present DINO (\textbf{D}ETR with \textbf{I}mproved de\textbf{N}oising anch\textbf{O}r boxes), a state-of-the-art end-to-end object detector. % in this paper. improves over previous DETR-like models performance and efficiency by using contrastive way for denoising training, mixed query selection method anchor initialization, look forward twice scheme box prediction. achieves $49.4$AP $12$ epochs $51.3$AP $24$ on COCO ResNet-50 backbone multi-scale features, yielding significant improvement...

We propose a novel regularized adaptation technique for context dependent deep neural network hidden Markov models (CD-DNN-HMMs). The CD-DNN-HMM has large output layer and many layers, each with thousands of neurons. huge number parameters in the makes challenging task, esp. when set is small. developed this paper adapts model conservatively by forcing senone distribution estimated from adapted to be close that unadapted model. This constraint realized adding Kullback-Leibler divergence...

Mining subtopics from weblogs and analyzing their spatiotemporal patterns have applications in multiple domains. In this paper, we define the novel problem of mining theme propose a probabilistic approach to model subtopic themes simultaneously. The proposed discovers by (1) extracting common weblogs; (2) generating life cycles for each given location; (3) snapshots time period. Evolution can be discovered comparative analysis snapshots. Experiments on three different data sets show that...

A longstanding question in computer vision concerns the representation of 3D shapes for recognition: should be represented with descriptors operating on their native formats, such as voxel grid or polygon mesh, can they effectively view-based descriptors? We address this context learning to recognize from a collection rendered views 2D images. first present standard CNN architecture trained shapes' independently each other, and show that shape recognized even single view at an accuracy far...

Convolutions are the fundamental building blocks of CNNs. The fact that their weights spatially shared is one main reasons for widespread use, but it also a major limitation, as makes convolutions content-agnostic. We propose pixel-adaptive convolution (PAC) operation, simple yet effective modification standard convolutions, in which filter multiplied with varying kernel depends on learnable, local pixel features. PAC generalization several popular filtering techniques and thus can be used...

We present in this paper a novel query formulation using dynamic anchor boxes for DETR (DEtection TRansformer) and offer deeper understanding of the role queries DETR. This new directly uses box coordinates as Transformer decoders dynamically updates them layer-by-layer. Using not only helps explicit positional priors to improve query-to-feature similarity eliminate slow training convergence issue DETR, but also allows us modulate attention map width height information. Such design makes it...

An improved human-robot collaborative control scheme is proposed in a teleoperated minimally invasive surgery scenario, based on hierarchical operational space formulation of seven-degree-of-freedom redundant robot. Redundancy exploited to guarantee remote center motion (RCM) constraint and provide compliant behavior for the medical staff. Based implemented framework, an RCM safe are applied nullspace achieve surgical tasks with interaction. Due physical interactions, safety accuracy may be...

This paper studies the optimal distribution of feet forces and control multilegged robots with uncertainties in both kinematics dynamics. First, a constrained dynamics for environment model are established by considering kinematic dynamic uncertainties. Under an external wrench robots, foot moments supporting legs can be formulated as quadratic programming problems subject to linear nonlinear constraints. The neurodynamics recurrent neural network is developed force optimization. For...

Accurate path tracking and stability are the main challenges of lateral motion control in mobile robots, especially under situation with complex road conditions. The interaction force between robots external environment may cause interference, which should be considered to guarantee its performance dynamic uncertain environments. In this article, a flexible scheme is for developed wheel-legged robot, consists cubature Kalman algorithm evaluate centroid slip angle yaw rate. Furthermore, fuzzy...

In this paper, we present an open-set object detector, called Grounding DINO, by marrying Transformer-based detector DINO with grounded pre-training, which can detect arbitrary objects human inputs such as category names or referring expressions. The key solution of detection is introducing language to a closed-set for concept generalization. To effectively fuse and vision modalities, conceptually divide into three phases propose tight fusion solution, includes feature enhancer,...

Learning manipulation skills from open surgery provides more flexible access to the organ targets in abdomen cavity and this could make surgical robot working a highly intelligent friendly manner. Teaching by demonstration (TbD) is capable of transferring human humanoid robots employing active learning multiple demonstrated tasks. This work aims transfer motion demonstrations manipulators robot-assisted minimally invasive (RA-MIS) using TbD. However, kinematic constraint should be respected...

For bilateral teleoperation, the haptic feedback demands availability of accurate force information transmitted from remote site. Nevertheless, due to limitation size, sensor is usually attached outside patient's abdominal cavity for surgical operation. Hence, it measures not only interaction forces on tip but also tool dynamics. In this letter, a model-free based deep convolutional neural network (DCNN) structure proposed dynamics identification, which features fast computation and noise...

Human activity recognition (HAR) using smartphones provides significant healthcare guidance for telemedicine and long-term treatment. Machine learning deep (DL) techniques are widely utilized the scientific study of statistical models human behaviors. However, performance existing HAR platforms is limited by complex physical activity. In this article, we proposed an adaptive real-time monitoring system activities (Ada-HAR), which expected to identify more motions in dynamic situations. The...

In the development of digital agriculture, agricultural robots play a unique role and confer numerous advantages in farming production. From invention first industrial 1950s, have begun to capture attention both research industry. Thanks recent advancements computer science, sensing, control approaches, experienced rapid evolution, relying on various cutting-edge technologies for different application scenarios. Indeed, significant refinements been achieved by integrating perception,...

Many previous works of soft wearable exoskeletons (exosuit) target at improving the human locomotion assistance, without considering impedance adaption to interact with unpredictable dynamics and external environment, preferably outside laboratory environments. This article proposes a novel hierarchical human-in-the-loop paradigm that aims produce suitable assistance powers for cable-driven lower limb exosuits aid ankle joint in pushing off ground. It includes two primary loop layers:...

In next-generation network architecture, the Cybertwin drove sixth generation of cellular networks sixth-generation (6G) to play an active role in many applications, such as healthcare and computer vision. Although previous (5G) provides concept edge cloud core cloud, internal communication mechanism has not been explained with a specific application. This article introduces possible based multimodal (beyond 5G) for electrocardiogram (ECG) patterns monitoring during daily activity. paradigm...

Score distillation sampling (SDS) has shown great promise in text-to-3D generation by distilling pretrained large-scale text-to-image diffusion models, but suffers from over-saturation, over-smoothing, and low-diversity problems. In this work, we propose to model the 3D parameter as a random variable instead of constant SDS present variational score (VSD), principled particle-based framework explain address aforementioned issues generation. We show that is special case VSD leads poor samples...

The presence of unknown physical interaction between the patients’ body and surgical tool in laparoscopic surgery requires a secure end-effector positioning while assuring reliable constraint motion. In this work, task-space control approach based on fuzzy approximation is proposed for teleoperated scenario utilizing serial redundant robot manipulator (7 degrees freedom), motions which are constrained with respect to point known as remote center motion (RCM). dynamical uncertainties due...

In the field of robotics, soft robots have been showing great potential in areas medical care, education, service, rescue, exploration, detection, and wearable devices due to their inherently high flexibility, good compliance, excellent adaptability, natural safe interactivity. Pneumatic occupy an essential position among because features such as lightweight, efficiency, non-pollution, environmental adaptability. Thanks its mentioned benefits, increasing research interests attracted...

Playing games between humans and robots have become a widespread human-robot confrontation (HRC) application. Although many approaches were proposed to enhance the tracking accuracy by combining different information, problems of intelligence degree robot anti-interference ability motion capture system still need be solved. In this paper, we present an adaptive reinforcement learning (RL) based multimodal data fusion (AdaRL-MDF) framework teaching hand play Rock-Paper-Scissors (RPS) game...