RFID technology has recently been exploited for not only identification but also sensing including trajectory tracking and gesture recognition. While contact-based (an tag is attached to the target of interest) achieved promising results, contactless still faces severe challenges such as low accuracy situation gets even worse when non-static, restricting its applicability in real world deployment. In this work, we present TagRay, a RFID-based system, which significantly improves accuracy,...

Two challenges in computer vision (CV) related to face detection are the difficulty of acquisition target domain and degradation image quality. Especially low-light situations, poor visibility images is difficult label, which results detectors trained under well-lit conditions exhibiting reduced performance environments. Conventional works enhancement object techniques unable resolve inherent difficulties collecting labeling images. The Dark-Illuminated Network with Contrastive...

Human-robot collaboration plays an important role in intelligent manufacturing. However, the main challenge is how robot can make online reactive changes to plan based on observed human behavior ensure completion of user-defined tasks. Such a further exacerbated if eye-in-hand manipulation considered since local field camera view cannot capture global observations. Different from existing planning approaches that separate perception and modules, strong assumptions about abilities, we develop...

The motion and large environmental loads experienced by riser cables connected to floating offshore wind turbines put them at higher risk of failure than grounded sections cable. We propose an Autonomous Riser Inspection System (ARIS) facilitate regular inspection for early fault detection gathering engineering data improve future cable design. Novel robotic methods automatic attachment, traversal are described. develop the sensing intelligent processing needed (i) enable autonomous position...

This work develops a fast mission planning framework named decision tree (PDT), that can handle large-scale multi-robot systems with temporal logic specifications in real time. Specifically, PDT builds incrementally to represent the task progress. The system states are modeled by both completion positions and times, which avoids sophisticated product automaton significantly reduces search space. By growing from root node leaf nodes, be searched for plannings satisfy linear (LTL) task....

It is challenging to coordinately allocate and plan the tasks of a heterogeneous multi-agent system in shared workspace. can be even more if agents are subject limited communication capability (i.e., exchange information with nearby only) complex temporal logic constraints. Motivated by these practical challenges, distributed task allocation planning method developed, which each agent communicates neighboring about preference sub-tasks executed estimated completion time) predicts out range....

This work develops a fast task allocation framework for heterogeneous multi-robot systems subject to both temporal logic and inter-task constraints. The considered constraints include unrelated tasks, compatible exclusive tasks. To specify such relationships, we extend conventional atomic proposition batch propositions, which gives rise the LTL <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^{\mathcal...

Temporal logic-based motion planning has been extensively studied to address complex robotic tasks. However, existing works primarily focus on static environments or assume the robot full observations of environment. This limits their practical applications since real-world are often dynamic, and robots may suffer from partial observations. To tackle these issues, this study proposes a framework for vision-based reactive temporal logic (V-RTLMP) integrated with LiDAR sensing. The V-RTLMP is...

Scribble-based weakly supervised segmentation techniques offer comparable performance to fully methods while significantly reducing annotation costs, making them an appealing alternative. Existing often rely on auxiliary tasks enforce semantic consistency and use hard pseudo labels for supervision. However, these overlook the unique requirements of models trained with sparse annotations. Since model must predict pixel-wise maps limited annotations, ability handle varying levels richness is...

Distribution shifts widely exist in medical images acquired from different centers, hindering the deployment of semantic segmentation models trained on data one center (source domain) to another (target domain). While unsupervised domain adaptation (UDA) has shown significant promise mitigating these shifts, it poses privacy risks due sharing between centers. To facilitate while preserving privacy, source-free (SFDA) and test-time (TTA) have emerged as effective paradigms, relying solely...

Although recent years have witnessed significant advancements in medical image segmentation, the pervasive issue of domain shift among images from diverse centres hinders effective deployment pre-trained models. Many Test-time Adaptation (TTA) methods been proposed to address this by fine-tuning models with test data during inference. These methods, however, often suffer less-satisfactory optimization due suboptimal direction (dictated gradient) and fixed step-size (predicated on learning...

In this work, we devise RF-DJ, a contactless music recognition system with the help of COTS RFID device. Since is caused by vibration and can influence RF signal, our could accurately recover frequency every tone, especially string instruments. Specifically, RF-DJ immune to noises from player/instrument motions ambient environment. Further more, it high signal relatively low sampling rate data. demonstration, put one tag on surface ukulele (not string) achieve overall accuracy $93%, 90%,...

In this study, we propose a novel top-down grasping approach for robots that combines deep high resolution convolutional neural network (DHRNet) and multiview perception-based trajectory planning controller (MP-PC). Unlike the traditional encoder-decoder architecture, DHRNet preserves high-resolution feature maps integrates at different scales to ensure maximum retention of spatial information its fusion with high-level semantic information. The MP-PC continuously adapts robotic end-effector...

Learning-based policy optimization methods have shown great potential for building general-purpose control systems. However, existing still struggle to achieve complex task objectives while ensuring safety during learning and execution phases black-box To address these challenges, we develop data-driven safe (D <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$^{2}$</tex-math> </inline-formula> SPO), a novel...

In resent years, providing location services for mobile targets in a closed environment has been growing interest. order to provide good localization and tracking performance drones GPS-denied scenarios, this paper proposes multi-tag radio frequency identification (RFID) system that is easy equip does not take up the limited resources of drone which susceptible processor cost constraints compared with computer vision based approaches. The passive RFID tags, no battery equipped, have an...