Time-Of-Flight (ToF) sensors efficiently capture scene depth, but the nonlinear depth construction procedure often results in extremely large noise variance or even invalid areas. Recent methods based on deep neural networks (DNNs) achieve enhanced ToF denoising accuracy but tend to struggle when presented with severe noise corruption due to limited prior knowledge of ToF data distribution. In this paper, we propose DepthCAD, a novel ToF denoising approach that ensures global structural smoothness by leveraging the rich prior knowledge in Stable Diffusion and maintains local metric accuracy by steering the diffusion process with confidence guidance. To adopt the pretrained image diffusion model to ToF depth denoising, we apply the diffusion on raw ToF correlation measurements with dynamic range normalization before converting to depth maps. Experimental results validate the state-of-the-art performance of the proposed scheme, and the evaluation on real data further verifies its robustness against real-world ToF noise.

Multimodal large language models (MLLMs) have achieved significant progress in image and language tasks due to the strong reasoning capability of large language models (LLMs). Nevertheless, most MLLMs suffer from limited spatial reasoning ability to interpret and infer spatial arrangements in three-dimensional space. In this work, we propose a novel vision encoder based on hierarchical fusion of geometry and semantics features, generating spatial-aware visual embedding and boosting the spatial grounding capability of MLLMs. Specifically, we first unveil that the spatial ambiguity shortcoming stems from the lossy embedding of the vision encoder utilized in most existing MLLMs (e.g., CLIP), restricted to instance-level semantic features. This motivates us to complement CLIP with the geometry features from vision-only self-supervised learning via a hierarchical adapter, enhancing the spatial awareness in the proposed SpatialGeo. The network is efficiently trained using pretrained LLaVA model and optimized with random feature dropping to avoid trivial solutions relying solely on the CLIP encoder. Experimental results show that SpatialGeo improves the accuracy in spatial reasoning tasks, enhancing state-of-the-art models by at least 8.0% in SpatialRGPT-Bench with ∼50% less memory cost during inference. The source code is available via https://ricky-plus.github.io/SpatialGeoPages/.

As a critical branch of the Internet of Things (IoT) in the medicine industry, the Internet of Medical Things (IoMT) significantly improves the quality of healthcare due to its real-time monitoring and low medical cost. Benefiting from edge and cloud computing, IoMT is provided with more computing and storage resources near the terminal to meet the low-delay requirements of computation-intensive services. However, the service offloading from health monitoring units (HMUs) to edge servers generates additional energy consumption. Fortunately, artificial intelligence (AI), which has developed rapidly in recent years, has proved effective in some resource allocation applications. Taking both energy consumption and delay into account, we propose an energy-aware service offloading algorithm under an end-edge-cloud collaborative IoMT system with Asynchronous Advantage Actor-critic (A3C), named ECAC. Technically, ECAC uses the structural similarity between the natural distributed IoMT system and A3C, whose parameters are asynchronously updated. Besides, due to the typical delay-sensitivity mechanism and time-energy correction, ECAC can adjust dynamically to the diverse service types and system requirements. Finally, the effectiveness of ECAC for IoMT is proved on real data.

This paper presents an active control method, consisting of an adaptive sliding-mode controller (ASMC) and a displacement field reconstruction (DFR) method, for chatter suppression in turning of thin-walled workpieces (such as compressor disks and casings in aircraft engines) where low workpiece stiffness renders machining with potential regenerative chatter. Due to the presence of multi-modal dynamics, variant modal parameters, and measurement difficulties, active chatter control of thin plate turning has been challenging. Unlike existing controls based on a lumped-parameter single degree-of-freedom cutting model, a distributed-parameter dynamic model of a rotating thin plate with multiple vibration modes is used to analyze the machining stability with the designed controller. Moreover, model parameters of the plate are not needed to construct the controller. The DFR is employed to capture the plate dynamic behavior for feedback to the ASMC during turning, overcoming the long existing difficulties to measure plate vibration at the cutting point. A fast tool servo is utilized in the control implementation. Theoretical analyses, numerical simulations, and experimental evaluation on a lathe demonstrate that chatter in thin plate turning can be effectively attenuated with the proposed active control method.

Motivated by the needs to develop dexterous flexonic mobile node (FMN) for structural health monitoring where obstacle avoidance is among the most challenging tasks, this paper presents a novel FMN design for navigating on iron-based structures. Unlike most existing magnetic wheel designs, the FMN uses fixed-magnet configuration contributing to smooth motion, self-maintained orientation and automatic negotiation of concave corners. Taking into account of several stringent considerations (including tight dimension, efficient attachment and manipulation flexibility) to facilitate avoidance of structural obstacles in tight space, a prototype FMN incorporating a compliant beam (that results in more DOFs than designs with multiple links and pin/ball joints), along with a 3-step strategy for negotiating a convex corner, has been developed. Although controlling a continuous deforming beam is challenging given limited number of actuators, both simulation and experimental results confirm a simple input/output relation between the rear axle displacement and front axle rotation, which is anticipated to facilitate control implementation of FMN for obstacle avoidance.

This paper presents a dynamic model of a knee joint interacting with a two-link exoskeleton for investigating the effects of different exoskeleton designs on internal joint forces. The closed kinematic chain of the leg and exoskeleton has a significant effect on the joint forces in the knee. A bio-joint model is used to capture this effect by relaxing a commonly made assumption that approximates a knee joint as a perfect engineering pin-joint in exoskeleton design. Based on the knowledge of a knee-joint kinematics, an adaptive knee-joint exoskeleton has been designed by incorporating different kinematic components (such as a pin, slider and cam profile). This design potentially eliminates the negative effects associated with the closed leg/exoskeleton kinematic chain on a human knee. An investigation in the flexion motion of an artificial human knee joint is presented to compare performances of five exoskeleton designs against the case with no exoskeletons. Analytical results that estimate internal forces using the dynamic model (based on the properties of a knee joint) agree well with the experiments. These studies lead to an adaptive mechanism with a slider/cam as an alternative to pin joints for the exoskeleton, and illustrate the application of the model for designing an adaptive mechanism that minimizes internal joint forces due to a human-exoskeleton interaction.