This paper proposes a novel modular selfreconfigurable robot (MSRR) "FreeBOT", which can be connected freely at any point on other robots. FreeBOT is mainly composed of two parts: spherical ferromagnetic shell and an internal magnet. The connection between the modules genderless instant, since magnet attract shells, not need to precisely aligned with specified connector. method has fewer physical constraints, so system extended more configurations meet functional requirements. accomplish...

The RGB-Thermal (RGB-T) information for semantic segmentation has been extensively explored in recent years. However, most existing RGB-T usually compromises spatial resolution to achieve real-time inference speed, which leads poor performance. To better extract detail information, we propose a two-stage Feature-Enhanced Attention Network (FEANet) the task. Specifically, introduce Module (FEAM) excavate and enhance multi-level features from both channel views. Benefited proposed FEAM module,...

Depth completion aims at predicting dense pixel-wise depth from an extremely sparse map captured a sensor, e.g., LiDARs. It plays essential role in various applications such as autonomous driving, 3D reconstruction, augmented reality, and robot navigation. Recent successes on the task have been demonstrated dominated by deep learning based solutions. In this article, for first time, we provide comprehensive literature review that helps readers better grasp research trends clearly understand...

Recently, RGB-Thermal based perception has shown significant advances. Thermal information provides useful clues when visual cameras suffer from poor lighting conditions, such as low light and fog. However, how to effectively fuse RGB images thermal data remains an open challenge. Previous works involve naive fusion strategies merging them at the input, concatenating multi-modality features inside models, or applying attention each modality. These are straightforward yet insufficient. In...

This paper proposed a novel tree climbing robot "Treebot" that has high maneuverability on an irregular environment and surpasses the state of art robots. Treebot's body is continuum maneuver structure degrees freedom superior extension ability. Treebot also equips with pair omni-directional grippers enable to adhere wide variety trees range gripping curvature. By combining these two designs, able reach many places including branches. can complex environment, but only five actuators are used...

Coverage path planning is a major application for mobile robots, which requires robots to move along planned cover the entire map. For large-scale tasks, multirobot systems offer significant advantages. In this letter, we propose Turn-minimizing Multirobot Spanning Tree Star (TMSTC*), an improved coverage (mCPP) algorithm based on MSTC*. Our partitions map into minimum bricks as tree branches, transforming problem finding maximum independent set of bipartite graph. We then use greedy...

Abstract Terrestrial self-reconfigurable robot swarms offer adaptable solutions for various tasks. However, most existing are limited to controlled indoor settings, and often compromise stability due their freeform connections. To address these issues, we present a snail robotic swarm system inspired by land snails, tailored unstructured environments. Our also employs two-mode connection mechanism, drawing from the adhesive capabilities of snails. The free mode, mirroring snail’s natural...

In this paper, a novel soft inflatable arm is proposed for telepresence robots. The new structure of the achieved by very common and low-cost material it light, weighing only about 50 g. However, can realize agile movement driving six tiny cables installed in shoulder elbow joints. work pumping air at low pressure (7.32 ± 3.45 kPa) allows direct human contact without any external force sensors. This paper proposes joint compressed models, kinematic models whole robot arm. These be easily...

This paper proposes a novel freeform strut-node structured modular self-reconfigurable robot (MSRR) called FreeSN, consisting of strut and node modules. A module is mainly low-carbon steel spherical shell. contains two connectors, which provide strong magnetic connections flexible motions. The FreeSN system shares the benefits connection structures. brings good adaptability to environment. triangle substructures inside configuration significantly improve structural stability. parallel...

The width of a neural network matters since increasing the will necessarily increase model capacity. However, performance does not improve linearly with and soon gets saturated. In this case, we argue that number networks (ensemble) can achieve better accuracy-efficiency trade-offs than purely width. To prove it, one large is divided into several small ones regarding its parameters regularization components. Each these has fraction original one's parameters. We then train together make them...

This paper proposes a novel modular self-assembling, self-reconfiguring robot with the 3D continuous dock called "SnailBot". SnailBot mainly consists of spherical ferromagnetic shell and six-wheel rocker chassis embedded magnets. Unlike many other existing self-reconfigurable robots fixed docking locations, uses to attach its peers regardless alignment. freeform mechanism can greatly improve efficiency self-reconfiguration reduce failures because there is nearly no constraint in location...

Aiming to develop a more robust and intelligent heterogeneous system for adversarial catching in security rescue tasks, this article, we discuss the specialities of applying asymmetric self-play curriculum learning techniques deal with increasing heterogeneity number different robots modern multirobot systems (HMRS). Our method, based on actor-critic multiagent reinforcement learning, provides framework that can enable cooperative behaviors among teams. This leads development an HMRS complex...

This paper presents an omnidirectional steer-by-wire system for a four-wheel independent steering vehicle. The consists of extended interface and behavior-based controller. provides novel manipulation way to vehicle driver. driver can control the in traditional or omnidirectionally without any mode switching operation. reservation driving makes adapt easily. also has force-feedback function synchronize orientations wheels so as improve handling. controller is designed controlling wheels. It...

In this paper, we propose an autonomous tree climbing strategy for a novel robot that is named Treebot. The proposed algorithm aims to guide Treebot in along optimal path by the use of minimal sensing resources. Inspired inchworms, reconstructs shape simply tactile sensors. It reveals how realization environment can be achieved with limited information. An efficient nonholonomic motion planning also make climb on path. This accomplished prediction future tree. study presented paper includes...

Abstract Treebot is the first tree‐climbing robot that capable of climbing from a tree trunk to branch. The employs several design principles adapted arboreal animals, including claw gripping and inchworm locomotion, with certain artificial optimization achieve high maneuverability on irregular‐shaped trees. composed pair grippers permits attach wide variety trees range curvature, novel continuum maneuvering structure provides adaptability. In actuation, only five actuators are necessary....

As vision based perception methods are usually built on the normal light assumption, there will be a serious safety issue when deploying them into low environments. Recently, deep learning have been proposed to enhance images by penalizing pixel-wise loss of and images. However, most suffer from following problems: 1) need pairs for training, 2) poor performance dark images, 3) amplification noise. To alleviate these problems, in this letter, we propose two-stage unsupervised method that...

In this work, we present a method that enables mobile robot to hand over objects humans efficiently and safely by combining navigation with visual perception. Our robotic system can map its environment in real time locate pick up. It uses advanced algorithms grasp way suits human preference employs path planning obstacle avoidance navigate back the user. The adjusts movements during handover analyzing human’s posture through sensors, ensuring smooth collision-free handover. Tests of our show...

The intrinsic modularity and reconfigurability of modular reconfigurable robot (MRR) systems confer advantages such as versatility, fault tolerance, economic efficacy, thereby showcasing considerable potential across diverse applications. continuous evolution the technology landscape emergence conceptual designs have generated multiple MRR categories, each described by its respective morphology or capability characteristics, leading to some ambiguity in taxonomy. This paper conducts a...

In this paper, a novel soft inflatable arm is proposed for telepresence robots. It capable of imitating human arms to realize remote interaction. The new using very common and low cost material, it light, which weight only about 50 grams, but can well agile movement by driving three tiny cables installed in shoulder joint elbow joint, respectively. Meanwhile, the cable driven mechanism also allows connecting numbers joints easily. work just pumping air with pressure (7.32 ± 3.45 kPa),...

Accurate perception of the surrounding scene is helpful for robots to make reasonable judgments and behaviours. Therefore, developing effective representation recognition methods are significant importance in robotics. Currently, a large body research focuses on novel auxiliary features networks improve indoor ability. However, few them focus directly constructing object relations recognition. In this paper, we analyze weaknesses current propose an Object-to-Scene (OTS) method, which...

Audio-Visual scene understanding is a challenging problem due to the unstructured spatial-temporal relations that exist in audio signals and spatial layouts of different objects visual images. Recently, many studies have focused on abstracting features from convolutional neural networks while learning explicit semantically relevant frames sound images has been overlooked. To this end, we present an end-to-end framework, namely attentional graph network (AGCN), for structure-aware...

Abstract This article aims to improve the deep-learning-based surface defect recognition. In actual manufacturing processes, there are issues such as data imbalance, insufficient diversity, and poor quality of augmented in collected image for product A novel generation method with multiple loss functions, DG2GAN is presented this paper. employs cycle consistency generate images from a large number defect-free images, overcoming issue imbalanced original training data. DJS optimized...