In recent years, convolutional neural network (CNN) based methods have achieved great success in a large number of applications and been among the most powerful widely used techniques computer vision. However, CNN-based are com-putational-intensive resource-consuming, thus hard to be integrated into embedded systems such as smart phones, glasses, robots. FPGA is one promising platforms for accelerating CNN, but limited bandwidth on-chip memory size limit performance accelerator CNN.

Convolutional neural network (CNN) has become a successful algorithm in the region of artificial intelligence and strong candidate for many computer vision algorithms. But computation complexity CNN is much higher than traditional With help GPU acceleration, CNN-based applications are widely deployed servers. However, embedded platforms, solutions still too complex to be applied. Various dedicated hardware designs on field-programmable gate arrays (FPGAs) have been carried out accelerate...

Recent research on neural networks has shown a significant advantage in machine learning over traditional algorithms based handcrafted features and models. Neural are now widely adopted regions like image, speech, video recognition. But the high computation storage complexity of network inference poses great difficulty its application. It is difficult for CPU platforms to offer enough capacity. GPU first choice processes because capacity easy-to-use development frameworks. However,...

Recent researches on neural network have shown significant advantage in machine learning over traditional algorithms based handcrafted features and models. Neural is now widely adopted regions like image, speech video recognition. But the high computation storage complexity of inference poses great difficulty its application. CPU platforms are hard to offer enough capacity. GPU first choice for process because capacity easy use development frameworks. On other hand, FPGA-based accelerator...

Nowadays, wearable devices derived from flexible conductive hydrogels have attracted enormous attention. Nevertheless, the utilization of in practical applications under extreme conditions remains a significant challenge. Herein, series inorganic salt-ion-enhanced (HPE-LiCl) consisting hydroxyethyl cellulose, acrylate, lithium chloride, and ethylene glycol/water binary solvent were fabricated via facile one-pot method. Apart outstanding self-adhesion, high stretchability, remarkable fatigue...

Collaborative exploration in an unknown environment without external positioning under limited communication is essential task for multi-robot applications. For inter-robot positioning, various Distributed Simultaneous Localization and Mapping (DSLAM) systems share the Place Recognition (PR) descriptors sensor data to estimate relative pose between robots merge robots' maps. As maps are constantly shared among exploration, we design a map-based DSLAM framework, which only shares submaps,...

Multi-robot exploration in unknown environments is a fundamental task for multi-robot system, involving inter-robot communication through messages among the robots. However, restricted environment, limited resources become system's bottleneck due to large amount of data occupancy grid map. Hence, enhance multi-agent communication-constrained environments, this letter develops method build topological maps while robot moves environment and an strategy based on created The latter map comprises...

Pose estimation of object is one the key problems for automatic-grasping task robotics. In this paper, we present a new vision-based robotic grasping system, which can not only recognize different objects but also estimate their poses by using deep learning model, finally grasp them and move to predefined destination. The model demonstrates strong power in hierarchical features greatly facilitates recognition mission. We apply Max-pooling Convolutional Neural Network (MPCNN), most popular...

Recently, Deep Learning (DL), especially Convolutional Neural Network (CNN), develops rapidly and is applied to many tasks, such as image classification, face recognition, segmentation, human detection. Due its superior performance, DL-based models have a wide range of application in areas, some which are extremely safety-critical, e.g. intelligent surveillance autonomous driving. the latency privacy problem cloud computing, embedded accelerators popular these safety-critical areas. However,...

Geese are among the most important poultry species in world. The current generally accepted hypothesis is that European domestic geese originated from greylag (Anser anser), and Chinese have two origins, of which swan cygnoides), Yili goose geese. To explain origin demographic history geese, we selected 14 breeds Europe China wild populations whole genome sequencing data were obtained for 74 samples.Population structure analysis phylogenetic trees showed ancestor except Yili, Analysis...

This paper presents SANA-1.5, a linear Diffusion Transformer for efficient scaling in text-to-image generation. Building upon SANA-1.0, we introduce three key innovations: (1) Efficient Training Scaling: A depth-growth paradigm that enables from 1.6B to 4.8B parameters with significantly reduced computational resources, combined memory-efficient 8-bit optimizer. (2) Model Depth Pruning: block importance analysis technique model compression arbitrary sizes minimal quality loss. (3)...

Feature-point extraction is a fundamental step in many applications, such as image matching and Simultaneous Localization Mapping (SLAM). The CNN-based feature-point methods have made significant signs of progress both detection descriptor generation compared with handcrafted processes. However, the computational storage complexity makes it difficult for CNN to run on real-time embedded systems. In this paper, we aim deploy advanced onto FPGA We optimize softmax data flow so that computation...

Planes, lines, and cylinders widely exist in man-made environments. This letter introduces a LiDAR simultaneous localization mapping (SLAM) system using those three types of landmarks. Our algorithm has components including local mapping, global localization. The jointly adjust planes, with poses to minimize the point-to-model cost, which is referred as plane-line-cylinder adjustment (PLCA). We prove that, some preprocessing, PLCA independent number points captured from landmarks, makes...

Autonomous exploration and mapping of unknown terrains employing single or multiple robots is an essential task in mobile robotics has therefore been widely investigated. Nevertheless, given the lack unified data sets, metrics, platforms to evaluate approaches, we develop autonomous robot benchmark en-titled Explore-Bench. The involves various explo-ration scenarios presents two types quantitative metrics efficiency multi-robot cooperation. Explore-Bench extremely useful as, recently, deep...

Rubber concrete (RC) is a new type of that currently receiving lot attention, solving serious pollution problems by grinding waste tires into granules and adding them to concrete. However, rubber has deficiencies in mechanics durability, been reinforced fibers many studies. In this study, the mechanical durability properties with added polypropylene basalt (PBRC) were investigated series experiments including apparent morphology, mass, static compressive tensile tests, ultrasonic...

In recent years, Convolutional Neural Network (CNN) has been widely applied in computer vision tasks. FPGAs have explored to accelerate CNNs due its high performance, energy efficiency, and flexibility. By fusing multiple layers CNN, the intermediate data transfer can be reduced. With a faster algorithm using Winograd transformation, computation of convolution further accelerated. However, previous accelerators with cross-layer or are designed for particular CNN model. The FPGA should...

Disaster areas involving floods and earthquakes are commonly large, with the rescue time being quite tight, suggesting multi-Unmanned Aerial Vehicles (UAV) exploration rather than employing a single UAV. For such scenarios, current UAV is modeled as Coverage Path Planning (CPP) problem to achieve full area coverage in presence of obstacles. However, UAV's endurance capability limited, constrained, prohibiting even multiple UAVs from completing disaster on time. Therefore, this paper defines...

In recent years, Convolutional Neural Network (CNN) has been widely applied in computer vision tasks and achieved significant improvement image object detection. The CNN methods consume more computation as well storage, so GPU is introduced for real-time However, due to the high power consumption of GPU, it difficult adopt mobile applications like automatic driving. previous work proposes some optimizing techniques lower detection on or FPGA. first Low-Power Image Recognition Challenge...

Training convolutional neural network (CNN) usually requires large amount of computation resource, time and power. Researchers cloud service providers in this region needs fast efficient training system. GPU is currently the best candidate for CNN training. But FPGAs have already shown good performance energy efficiency as inference accelerators. In work, we design a compressed process together with an FPGA-based accelerator We adopt two widely used model compression methods, quantization...

In recent years, Convolutional Neural Networks (CNNs) have been widely applied in computer vision and achieved significant improvements object detection tasks. Although there are many optimizing methods to speed up CNN-based algorithms, it is still difficult deploy algorithms on real-time low-power systems. Field-Programmable Gate Array (FPGA) has explored as a platform for accelerating CNN due its promising performance, high energy efficiency, flexibility. Previous works show that the...

Collaborative perception in unknown environments is a critical task for multi-robot systems. Without external positioning, mapping systems have relied on the transfer of place recognition (PR) descriptors or sensor data relative pose estimation (RelPose) and share their local maps localization. Thus, communication limited environment, transmission can become significant bottleneck system. To address this limitation, we propose MR-GMMapping, Multi-Robot GMM-based system which robots perform...

The effects of rare earth erbium (Er) micro-additions on the microstructures and mechanical properties 2024 aluminum alloy were investigated. fracture surfaces specimens prepared via high-energy ball milling, cold isostatic pressing microwave sintering carried out by optical microscopy (OM) scanning electron (SEM). Under conditions heating rate 20 min/°C soaking time 30 min at 490 °C, it was found that with increase in Er addition, grain size first decreased then increased, reached a minimum...

Decentralized visual simultaneous localization and mapping (DSLAM) can share locations environmental information between robots, which is an essential task for many multi-robot applications. The odometry (VO) a basic component to estimate the 6-DoF absolute pose robot place recognition (DPR) fundamental element produce candidate matches sharing among different robots. goal of this paper build CNN-based real-time DSLAM system on embedded FPGA platforms. Because high precision requirement VO,...

In recent years, Convolutional Neural Network (CNN) has been widely used in robotics, which dramatically improved the perception and decision-making ability of robots. A series CNN accelerators have designed to implement energy-efficient on embedded systems. However, despite high energy efficiency accelerators, it is difficult for robotics developers use it. Since various functions robot are usually implemented independently by different developers, simultaneous access accelerator these...