Unmanned Aerial Vehicles (UAVs) play a crucial role in various scenarios ranging from disaster response to traffic surveillance. However, aerial video footage often suffers from severe motion blur due to rapid flight maneuvers, vibrations, and camera panning, which can significantly degrade downstream tasks such as target detection. Our goal is to explore a computationally-efficient and effective video deblurring approach to enhance UAV target detection performance. To reduce computational cost, we first propose an Adaptive Latent Scale Selector that dynamically adjusts the latent space resolution according to the intensity of UAV motion, thus balancing detail preservation with inference efficiency. To ensure temporal consistency, we introduce a Multi-Frame Alignment and Learnable Gating module to warp and gate the preceding frames, allowing the model to fuse only relevant temporal information and suppress misaligned or uninformative features. Our method can effectively recover sharp details from the UAV video stream. Extensive experiments on real UAV benchmarks demonstrate that our method not only yields superior deblurring performance but also significantly boosts target detection accuracy, making it highly applicable to robust aerial vision tasks. Code will be publicly available here.

Underwater images suffer from color distortion, haze, and poor visibility due to light refraction and absorption in water. These challenges significantly impact the utilization of Autonomous Underwater Vehicles (AUVs) or marine robots. Typically, color and brightness distortions manifest at lower frequencies, while edge and texture distortions are prevalent at higher frequencies. Traditional methods struggle to concurrently rectify these mixed distortions as they primarily concentrate on the spatial domain. To address these issues, we introduce the Dynamic SpectraFormer, which enhances under-water images through a frequency domain transformer. The Dynamic SpectraFormer introduces an ultra-high-resolution sparse spectrum attention module, which could capture the long-term dependency without losing the universal approximating power. Additionally, we have developed a dynamic spectrum weight generation layer that serves as an adaptive spectrum band selector, accentuating critical frequency bands and suppressing less relevant ones. Consequently, this method significantly improves underwater image quality by addressing both high-and low-frequency distortions. Our extensive ablation studies and comparative evaluations consolidate the Dynamic SpectraFormer’s efficacy across multiple underwater image enhancement benchmarks. The source code is available at https://github.com/arifence2024/DynamicSpectraFormer.git.

In daily life, we realize various complex tasks with our two upper limbs based on the so-called bimanual coordination phenomenon. A fundamental feature of bimanual coordination is the natural tendency to synchronize the motion of two upper limbs, resulting in some preferred patterns of interlimb coordination. In this study, based on the coarse-to-fine human-robot collaboration framework, we investigate the possibility of realizing human-robot interaction and collaboration for accurate manipulation utilizing voluntary bimanual coordination. The practical motivation of utilizing the voluntary bimanual coordination, which can be perceived as an indirect way of implementing interlimb transmission of force feedback information, is to avoid the bad effect on force feedback presentation due to the counterforce if the human-robot collaboration was realized in an unimanual manner. For experimental studies, we firstly evaluated the synchronous protocols in terms of in-phase and anti-phase between two arms. Based on the studied protocol, moving target tracking was then demonstrated as an application scenario of the proposed human-robot collaboration utilizing voluntary bimanual coordination.

In this study, human-robot interaction with multimodal perception interface combining human visual and haptic perception is introduced. In the proposed human-robot collaboration method, cognitive capabilities of human and accurate motion control capabilities of robot are integrated based on a coarse-to-fine strategy. Human operator is designated for coarse global motion under feedback interfaces utilizing human visual as well as haptic modalities. Simultaneously, fine local motion in an active manner without involving human intention-aware is realized by a robotic module with high-speed actuators and high-speed sensory feedback. Experiments demonstrated the effectiveness of the proposed method.

Human-robot collaboration taking both human and robot's advantages becomes very promising in recent years. In this study, we propose a new active assistant robotic system for human-robot collaboration with the aim of optimally combining the cognitive capabilities of human and accurate motion control capabilities of robot. Under our method, human operator is provided with pneumatic haptic feedback to guide the system for coarse global-motion. Fine local-motion in an active manner is realized by a dynamic compensation robotic module with high-speed visual feedback. Since the active assist behavior here is confined to local motion and is accordingly safe to human operator. Experiments of preliminary investigation for set-point positioning and contour tracing under human-robot collaboration were implemented. Experimental results verified the effectiveness of the proposed method.

In this paper, we propose a robotic assistance system for carrying out a highly accurate peg-in-hole task by using a high-speed vision system and high-accuracy actuators. In this system, the vision system measures the position and the posture of the workpieces, and the actuators are moved according to this information. By using the high-speed vision system and operating the system at 500 Hz, a hole formed in one workpiece can be made to follow the complicated motion of a peg operated by a human and can be accurately aligned with the peg. In addition, since the torque required to move the hole is calculated on the basis of the position and direction data of the workpieces in the coordinate plane of the image captured by the vision system, the system is not affected by camera installation errors. Moreover, by using two cameras and combining the information from them, we made the tracking field wider. In order to evaluate the performance of the system, we executed a peg-in-hole task using a 50μm peg and a 70μm hole, and showed that the system completed the task with a success rate of 90%.

This video introduces our recent studies on human-robot collaboration based on dynamic compensation framework with the aim of optimally combining the cognitive capabilities of human and accurate motion capabilities of robot. Under the dynamic compensation approach, human operator is for cognitive global-motion without caring much about accuracy. Fine local-motion in an active manner is realized by a dynamic compensation robotic module based on high-speed visual feedback. Application scenarios from micro-manipulation to macro-manipulation are implemented.

This paper focuses on high-performance robotic contour tracking under the uncertainties that commonly exist in actual robotic applications. These uncertainties can be attributed to the robot itself (such as modeling errors or mechanical defects like backlash) or to environmental issues (such as calibration errors or misalignment of the workpiece). We propose a non-model-based dynamic compensation approach based on the coarse-to-fine philosophy, which enables contour tracking with both high speed and good accuracy. This is achieved by adopting a methodology in which a main robot performs fast but coarse motion, while an add-on module conducts accurate compensation for the overall uncertainties using a high-speed camera and high-speed compensation actuator. An algorithm called pre-compensated proportionalderivative sliding mode control (pre-compensated PD-SMC) is proposed to control the compensation actuator. The effectiveness of the proposed contour tracking approach and control algorithm are experimentally verified using two typical planar-contour shapes: a random smooth-curvature and rectangle.

Realizing accurate robotic manipulation of flexible objects is very challenging. It asks for dexterous manipulation skill, and demands good adaptation to dynamic uncertainties due to deformation. Unlike traditional methods that basically depend on complex modeling, we follow the opposite approach by exploring possible simplifications based on high-speed motion strategy as well as high-speed visual sensing. As an application task concerned with robotic needle threading, this work demonstrates our philosophy by proposing a novel and straightforward high-speed strategy to speed up the manipulation while solves the issue of deformation uncertainty at the same time. The methodology is based on a very simple physical fact, namely that by rotating a thread with sufficient velocity, the thread can be approximated as a rigid object caused by the constantly applied centrifugal force during the rotation. Therefore, the needle threading problem is converted into a simpler peg-insertion problem, and the complexity of interaction between robot and deformable thread is significantly simplified. Based on the proposed strategy, simple technical methods then become possible for adoption to accomplish the challenging task. Experiments show that the proposed approach allows the needle threading task to be realized very rapidly.

This paper presents a visual compliance strategy to deal with the problem of fast peg-and-hole alignment with large position and attitude uncertainty. With the use of visual compliance and adoption of a light-weight 3-DOF active peg, decoupled alignment for position and attitude is realized. The active peg is capable of high-speed motion and with less dynamic defects than a traditional robot arm. Two high-speed cameras, one configured as eye-in-hand and the other as eye-to-hand are adopted to provide with the task-space feedback. Visual constraints for effecting the visual compliant motion are analyzed. Alignment experiments show that peg-and-hole alignment with the proposed approach could be successfully realized with robust convergence, and on average, the alignment could be realized within 0.7 s in our experimental setting.