In computer-assisted orthopedic surgery (CAOS), robust and accurate registration of the preoperative full bone model and the intraoperative partial point set is a prerequisite for reliable surgical navigation, yet remains highly challenging due to partial overlap, noise, and outliers. We propose the Overlap-Guided Bidirectional Hybrid Mixture Registration (OBHMR) framework for robust and accurate partial-to-full registration. First, geometric features (i.e., surface normals) extracted from raw point sets are incorporated in both correspondence estimation and transform computation. Meanwhile, we formulate a hybrid mixture model that jointly represents positions with Gaussian mixtures (GMMs) and normals with von Mises–Fisher (vMF) mixtures across the two generalized point sets. Second, a dual-branch overlap prediction network leverages feature similarity and geometric structure to provide accurate point-wise overlap scores that guide hybrid-mixture construction under partial overlap. Third, a correspondence module integrates rotation-invariant features, multi-level self-attention, and clustering-based refinement to enhance reliability under noise and misalignment. Finally, a bidirectional objective jointly aligns source-to-target and target-to-source mixtures, explicitly accounting for discrepancies induced by noise and outliers in both the preoperative and intraoperative point sets to achieve robust optimization. Extensive experiments on 1,399 femur and 1,301 hip models demonstrate superior performance over state-of-the-art methods across overlap ratios from 5% to 70%, under both isotropic and anisotropic noise and outlier ratios up to 100%, achieving errors as low as 1.27° rotation and 1.18 mm translation at 50% overlap with 2.5mm noise. Additional tests on liver and ModelNet40 confirm strong generalization across medical and non-medical data. Ablation studies further validate the contributions of normals, overlap estimation, and the bidirectional formulation.

This paper presents a novel robust and accurate normal-assisted learning-based rigid point set registration approach, i.e., Deep Bi-directional Hybrid Mixture Registration (DeepBHMR), where normal vectors are used in both correspondence and transformation computational stages while the bi-directional registration processes are considered. DeepBHMR consists of three components, (1) the correspondence estimation network that predicts the correspondence probabilities; (2) the posterior estimation module that computes the HMMs parameters; (3) the transformation estimation module that calculates the rigid transformation matrix by utilizing the bidirectional optimization mechanism. DeepBHMR has been extensively validated on various medical data sets, outperforming state-of-the-art registration methods. For femur bones, the mean rotation error value is approximately 1° (i.e, 1.01°) and the translation error is less than 1 mm (i.e., 0.30 mm) respectively, which meets the requirement of computer-assisted orthopedic surgery. Furthermore, even (1) trained with femur data and tested on distinct shapes and (2) under the large transformation, the mean RMSE values of registration are 2.60 mm and 3.05 mm respectively, demonstrating DeepBHMR’s favorable generalizability to different data shapes and great capability to handle global registration. Additionally, the individual significant contributions and computational efficiency of adopting normal vectors and utilizing the bidirectional mechanism have been validated in ablation studies. The results demonstrate the DeepBHMR’s favorable generalizability from femur bones to hip bones and that DeepBHMR can successfully handle the large transformation or partial-to-full registration simultaneously. The code implementation of DeepBHMR has been made publicly available at https://github.com/zzyrobot/DeepBiHMM.git.

Vehicle trajectory is crucial for autonomous driving. Relatively scattered trajectory data points pose difficulties in modeling the motion’s inherent continuity in spatial and temporal dimensions. Additionally, identifying the driving patterns of vehicles from trajectories is also a significant challenge. These implicit characteristic patterns are difficult to discern from the complex details of the trajectory data. To address these issues, we propose a new framework called Wavelet-Mamba Trajectory Prediction (WMTP), which fuses wavelet analysis through state-space modeling to capture global trends in driving patterns and details of vehicle motion. The approach employs the Discrete Wavelet Transform (DWT) to decompose trajectory data into wavelet coefficients in different time scales and frequencies, and then utilizes these coefficients to generate the trajectory through the Inverse Discrete Wavelet Transform (IDWT). An encoder-decoder neural architecture is proposed for learning potential temporal features from the input trajectory sequences, and these features are projected into the wavelet domain. Wavelet coefficients of future trajectories are generated using different scale-oriented decoders. The estimated coefficients are further used to realize the trajectory prediction via the IDWT module. Experiments demonstrate that WMTP exhibits excellent performance on three large-scale real-world trajectory prediction datasets, with promising robustness and inference speed. The research findings also verify the effectiveness of time - frequency analysis in trajectory prediction tasks.

Deep learning has been widely used in wearable sensors to improve accuracy in gait analysis. However, these deep learning models typically focus on single tasks, either in gait parameter estimation or gait phase detection. This study presents a novel multi-task learning framework for regression (i.e., gait cycle percentage prediction) and classification (i.e., gait phase prediction) tasks in pathological gait analysis using wearable sensors. Our framework employs a Multi-gate Mixture-of-Experts architecture to achieve soft parameter-sharing, integrating expert networks, cross-expert attention mechanisms, and dynamic routing to balance shared and task-specific representations. To reduce computational burden in wearable applications, we compare lightweight model configurations that optimize expert count and feature dimensionality. Model performance has been validated on a public dataset consisting of 158 frail older adults, demonstrating that our framework significantly outperforms single-task learning and hard parameter-sharing baselines, achieving an accuracy of 97.56% and a Mean Absolute Error (MAE) of 0.0397. Notably, the most compact lightweight configuration reduces the parameter count by nearly 98% (from 2.118 million to 0.0469 million), achieving an accuracy of 96.47% and a MAE of 0.0549. Attention mechanisms significantly enhance performance across all configurations, with improvements ranging from 17.9% to 30.4%. These findings validate the potential of lightweight multi-task approaches for real-time gait assessment, offering promising applications for clinical evaluation and rehabilitation monitoring in geriatric populations.

Robust and accurate 2D/3D registration, which aligns preoperative models with intraoperative images of the same anatomy, is crucial for successful interventional navigation. To mitigate the challenge of a limited field of view in single-image intraoperative scenarios, multi-view 2D/3D registration is required by leveraging multiple intraoperative images. In this paper, we propose a novel multi-view 2D/3D rigid registration approach comprising two stages. In the first stage, a combined loss function is designed, incorporating both the differences between predicted and ground-truth poses and the dissimilarities (e.g., normalized cross-correlation) between simulated and observed intraoperative images. More importantly, additional cross-view training loss terms are introduced for both pose and image losses to explicitly enforce cross-view constraints. In the second stage, test-time optimization is performed to refine the estimated poses from the coarse stage. Our method exploits the mutual constraints of multi-view projection poses to enhance the robustness of the registration process. The proposed framework achieves a mean target registration error (mTRE) of$0.79+2.17\ \mathbf{mm}$on six specimens from the DeepFluoro dataset, demonstrating superior performance compared to state-of-the-art registration algorithms.

In computer-assisted orthopedic surgery (CAOS), accurate point set registration is essential for enhancing surgical accuracy. However, the sparse and low-overlap nature of intraoperative point sets presents significant challenges for reliable registration. To deal with these challenges, we propose a novel registration-after-completion framework, where the intraoperative point set is first completed, after which the two full point sets are registered. Our main contributions include the follows. First, we propose a progressive two-stage strategy to progressively complete the sparse and partial intraoperative point set. Second, considering that 1) intra-operative point set contains noise 2) the point completion process is not perfect, and 3) the resolution of preoperative image is limited, we adopt the bidirectional hybrid mixture models (HMMs) to represent the point set pairs and formulate the probabilistic registration network. In the proposed novel correspondence network where a dual-path cross-attention mechanism is adopted for feature fusion and a clustering mechanism is leveraged for calculating point-to-mixture correspondences. Furthermore, the bidirectional registration mechanism is leveraged to compute the transformation based on estimated correspondences. Third, we have extensively validated the proposed approach on various datasets and bone phantoms. Our experiments on 1399 human femur and 1301 hip models demonstrate that our method achieves state-of-the-art performance across overlap rates from 15% to 35% and at various point counts (i.e., 25, 50, and 100 points) under conditions with less than 50% overlap. Additionally, real phantom experiments on femur and hip models validate the method’s performance in simulated surgical scenarios. Experiments on ModelNet40 further confirmed our method’s effectiveness and generalizability.

In this paper, we propose a novel unsupervised intraoperative liver deformation correction method, called Learning Coherent point drift Network (LCNet), for image-guided liver surgery (IGLS). We first estimate the correspondences between the preoperative and intraoperative point sets in the optimal transport (OT) module by leveraging both original points and extracted features. Afterwards, we compute the point-wise displacement vector by solving the involved matrix equation in the Transformation module, where the point localisation noise is explicitly considered and modeled. Additionally, we present three variants of the proposed approach, i.e., LCNet, LCNet-ED and LCNet-WD, where better registration performances of LCNet against the other two demonstrate the superiority of the utilised Chamfer loss. We have extensively evaluated LCNet on the MedShapeNet dataset consisting of 615 different liver shapes of real patients, and the 3Dircadb dataset comprising 20 liver models of real patients. Extensive experimental results under different deformation and noise magnitudes demonstrate that LCNet outperforms existing state-of-the-art registration algorithms and holds significant application potential in IGLS. For example, when the overlapping ratio between the preoperative and intraoperative point sets is 25%, the deformation magnitude is 8 mm, the maximum point localization noise magnitude is 2 mm and the rotation angle lies in the range of [−45°, 45°], LCNet achieves a root-mean-square error (RMSE) value being 3.21 mm on MedShapeNet dataset, significantly outperforming those of Lepard and RoITr being 5.41 mm (p < 0.001) and 4.90 mm (p < 0.001) respectively.

Quadrupeds can generate various motor behaviors with the muscle synergies activated by the central nervous system. However, versatile locomotion for quadruped robots remains challenging due to the complexity of the high-dimensional limb dynamics with many physical constraints. Current approaches typically apply a dedicated policy or controller for each motor behavior, which requires the optimization of a large number of parameters and training process is complicated. In this paper, we propose a Generalist Agent Learning Architecture (GALA) to learn diverse motor behaviors simultaneously with a single policy network for quadruped locomotion. GALA significantly decreases the number of trainable parameters while producing appropriate motor behaviors by simply reactivating the generalist policy based on different sensory feedback and commands at run time. We experimentally analyze and demonstrate the versatile locomotion delivered by GALA on both simulated and real quadruped robots in various environments. The code is available at https://github.com/vsislab/GALA.

Accurate trajectory training is a challenging issue of upper-limb rehabilitation robots. This paper presents a novel discrete-time neural network control method to address the problems of system uncertainties and tracking accuracy in repetitive trajectory training. This control method consists of both an adaptive neural network controller and a learning controller. The adaptive neural network controller satisfying persistent excitation condition enables not only stable tracking control, but also accurate learning for closed-loop system dynamics. The learning controller utilizes the learned knowledge to provide high-performance control. In order to examine the effectiveness of the proposed control method, a series of simulation and real-world experiments with system uncertainties were conducted, in comparison of proportion integration differentiation control, sliding mode control and event-triggered adaptive neural control. Results substantiate that the proposed control method can precisely learn the unknown dynamics of human-robot system along the subject-specific reference trajectories, and control the robot to assist the arm for accurate and fast trajectory tracking with small control gains by reutilizing the learned knowledge. This control method may play a role in accurate trajectory training for upper-limb rehabilitation robots. Note to Practitioners—This work is motivated by the practical requirements of rehabilitation robots in repetitive motor training. Trajectory tracking is a fundamental but efficient training mode of rehabilitation robots. However, uncertainty and nonlinearity of the human-robot system dynamics may increase the difficulty of controlling the robots for accurate, efficient and reliable trajectory tracking training. To this end, this paper proposes a learning-based control method, which could learn the uncertain and nonlinear system dynamics by utilizing an elaborately designed neural network controller and thus achieve superior control performance using the learned knowledge. This control method can be potentially applied in variety of rehabilitation robots, showing advantages for repetitive trajectory training. First, it can accurately mode the uncertain dynamics of human-robot system and achieve personalized rehabilitation. Second, it does not need any parameter adaptation in the similar repeated motions, and can be more easily designed with digital implementations, thereby achieving better performance in the aspects of time saving. Third, it can ensure the tracking accuracy of the rehabilitation robot for rehabilitation efficiency and avoid secondary injury.

The driving and steering capabilities of the six-wheeled independent robot enable it to adapt to complex terrain and perform a variety of tasks. But the traditional control method cannot fully utilize its flexibility of independent driving of each wheel, and the control effect is poor in complex environments. In this work , we propose a multi-wheel cooperative control method in a distributed framework. The method leverages the robot’s independent driving and steering capabilities, employing a distributed model predictive control framework to synchronize the motion of six wheels. This is achieved by formulating individual wheel predictive control objectives and establishing inter-wheel trajectory and synchronization constraints, enabling coordinated multi-wheel locomotion. Considering the unknown perturbation factors such as road slippage and rugged terrain in the actual environment, an error model-based extended state observer algorithm is designed as a feed-forward controller to estimate the unknown perturbations and generate the control volume gain matrix by combining the estimated values to improve the robot’s adaptability. The control method effectively solves the motion control problem of the robot in the complex environment and verifies its effectiveness and robustness in practical applications.

Traditional assembly tasks often require robots to transfer the acquired skills to new tasks. However, previous transfer reinforcement learning methods typically ignore the inherent relationship between the source and the target domain tasks. This requires a substantial amount of interaction data to compensate for this deficiency, and generally results in poor transfer effects. To address this issue, a strategy transfer method that establishes a shared feature space between the source domain and the target domain is proposed to enhance the efficiency of strategy learning on peg-in-hole assembly. Initially, by calculating the distance between each feature in the source and target domains, the features with small distance are selected as shared features. Subsequently, in order to determine the successful search state, this paper uses the jump state of contact force and the relative position between the peg and the hole as the judgment criterion. Lastly, search and insertion peg-in-hole assembly experiments are conducted to validate the generalization of the proposed strategy, demonstrating its capability to transfer from simulation to the real world.

Hindsight experience replay (HER) has greatly increased the possibility of using deep reinforcement learning (DRL) for robotic manipulation with sparse rewards. However, there are still concerns about low learning efficiency and poor performance due to its insufficient exploration ability and bias against the initial goal introduced by HER. In this article, to solve this problem, a multigoal robotic manipulation DRL method based on adaptive intrinsic curiosity and universal value network fitting (AIC-UVNF) is proposed to further improve the exploration ability and learning performance. Specifically, this method utilizes an improved curiosity mechanism to construct a joint intrinsic reward and adaptively adjust the proportion, which can enhance exploration ability and avoid excessive pursuit of novel states. In addition, a universal value network fitting approach is proposed to incorporate the initial goal into the value function fitting process, which employs the value of the initial goal to eliminate the bias of HER in the algorithm update. Combined with the off-policy soft actor-critic method, AIC-UVNF is verified on multigoal robotic manipulation tasks. The results show that the proposed method achieves better convergence efficiency and learning performance.

Despite significant advancements in multirobot technologies, efficiently and collaboratively exploring an unknown environment remains a major challenge. In this paper, we propose AIM-Mapping, an Asymmetric InforMation enhanced Mapping framework based on deep reinforcement learning. The framework fully leverages the privileged information to help construct the environmental representation as well as the supervised signal in an asymmetric actor-critic training framework. Specifically, privileged information is used to evaluate exploration performance through an asymmetric feature representation module and a mutual information evaluation module. The decision-making network employs the trained feature encoder to extract structural information of the environment and integrates it with a topological map constructed based on geometric distance. By leveraging this topological map representation, we apply topological graph matching to assign corresponding boundary points to each robot as long-term goal points. We conduct experiments in both iGibson simulation environments and real-world scenarios. The results demonstrate that the proposed method achieves significant performance improvements compared to existing approaches.

Real-time motion planning in dynamic environments presents a significant challenge for robotic manipulators. This paper introduces an innovative parallel Model Predictive Path Integral (MPPI) algorithm enabling the robot to navigate swiftly and safely in such environments. Unlike the conventional MPPI methods that rely on a single sequence of Gaussian means for trajectory sampling, the proposed Parallel MPPI (PMPPI) concurrently runs multiple planners with different strategies and adaptively integrates planned paths based on the current state, leveraging the advantages of different strategies and greatly improving the MPPI's exploration capability. Moreover, a Gradient-Velocity Modulated Signed Distance Field (SDF) cost function that dynamically adjusts costs based on the robot's velocity and the SDF gradient is defined, thereby promoting safer and purposeful motion planning. In the implementation, techniques like utilizing inverse kinematics solver for path guidance and Sparse Reward to expedite reaching time are integrated into the MPPI cost function design. Comparative evaluations against the traditional MPPI architecture and standard SDF cost designs demonstrate the superiority of the new method. Real-world experiments, including human-robot interaction, obstacle-crossing, and grasping tasks, validate the robustness and universality of our methodology, with average and maximum end effector speeds of 0.523 m/s and 1.225 m/s respectively.

Tower cranes are complex multi-input multioutput underactuated mechatronics systems. The anti-swing control issue of tower crane with varying suspension cable length and double spherical pendulum effect is still open. Furthermore, the system parameters uncertainty makes it more challenging to implement anti-swing control. In this study, we present an adaptive sliding mode anti-swing control approach based on time-delay estimation for underactuated tower crane with varying suspension cable length and double spherical pendulum effect. First, we employ the Lagrange’s method to develop a seven-degree-of-freedom (7-DOF) tower crane dynamic model that comprehensively accounts for jib slewing, trolley motion, payload hoisting/lowering, and payload/hook spherical swing within a three-dimensional (3-D) space. Then, a sliding mode surface is constructed by analyzing the nonlinear coupling relationship between the unactuated states and actuated states. The time-delay estimation technique with adaptive scheme can adapt and predicate unknown system parameters online. An adaptive sliding mode anti-swing control method with time-delay estimation is designed for 7-DOF tower crane system subject to the parameter uncertainties. The convergence of the closed-loop control system is carefully demonstrated through the Lyapunov stability theory. Finally, the hardware experiments verify the anti-swing control performance and robustness of the designed adaptive sliding mode controller. The superiority of the proposed adaptive sliding mode anti-swing controller is confirmed by a decrease of at least 42.09${\%}$and 58.33${\%}$in the maximum and residual payload swing, respectively, over state-of-the-art control methods.

In this paper, we introduce a novel overlap-based bidirectional point set registration approach, i.e., Overlap-guided Bidirectional Hybrid Mixture Registration (OBHMR), which incorporates geometric information (i.e., normal vectors) in both the correspondence and transformation stages and formulates the optimization objective of registration in a bidirectional manner. More importantly, to address the issue of partial-to-full registration, OBHMR utilises the predicted point-wise overlap score using networks to formulate the overlap-guided Hybrid Mixture Model consisting of the Gaussian Mixture Model (GMM) and Fisher Mixture Model (FMM). OBHMR contains four components: (1) the overlap-guided correspondence network that estimates the correspondence probabilities and calculates the point-wise overlap score; (2) the learning posterior module that estimates the overlap-guided HMM parameters; (3) the transformation module that computes the rigid transformation by formulating the optimisation objective in a bidirectional registration way, given correspondences and overlap-guided HMM parameters. Experiments using 1457 human femur and 1301 human hip models demonstrate significant improvements in partial-to-full registration performance (p < 0.01) under different overlapping ratios, compared to state-of-the-art registration approaches. Furthermore, individual contributions of three modules (i.e., additional normal vectors, overlap score estimation module and the bidirectional mechanism) in OBHMR have been validated in ablation studies. The results demonstrate OBHMR’s capability of tackling the challenging partial-to-full registration problems in computer-assisted orthopedic surgery. The codes are available at https://github.com/Dxinz/DeepOBHMR.

In this paper, we introduce a novel normal-assisted learning-based rigid registration approach, i.e., Deep Bi-directional Hybrid Mixture Registration (DeepBHMR). Our approach utilises helpful normal vectors explicitly in both correspondence and transformation stages and formulates the optimization objective of registration in a bi-directional way that considers noise in both point sets. DeepBHMR consists of three modules: (1) the correspondence network that estimates the correspondence probability relating points within one generalized point set (i.e., positional and normal vectors) with components of Hybrid Mixture Models (HMMs) representing the other generalized point set; (2) the posterior module that computes HMMs parameters; (3) the transformation module that computes the rotation matrix and the translation vector given the estimated generalized-point to hybrid-distribution correspondences and HMMs parameters. DeepBHMR has been validated on 291 human femur and 260 hip models, and extensive experimental results demonstrate that DeepBHMR outperforms the state-of-the-art registration methods (p-value < 0.01). In the circumstance of femur bones, the mean rotation and translation error values are around 1° (i.e., 1.01°) and less than 1 mm (i.e., 0.36mm), respectively. Furthermore, even under the large transformation (i.e., in the range of [0,180]° and [0, 100] mm), the mean RMSE values being 3.05 mm is still satisfactory. Additionally, the results demonstrate the DeepBHMR’s favorable generalizability from femur shapes to hip shapes. We have carefully validated the significant benefits of incorporating normal vectors and the bidirectional mechanism. DeepBHMR can successfully handle the challenging scenario of large transformation and partial registration. The codes are available at https://github.com/zzyrobot/DeepBHMR.git.

In this paper, we introduce Bidirectional Non-Overlapping Filtering Network (Bi-NOFNet), which registers the partial intraoperative point set with full preoperative point set for computer-assisted interventions (CAI). Our contributions are three-folds. First, Bi-NOFNet adopts customised feature extractor to extract distinctive features from both point sets, with which the per-point overlap mask is predicted and the overlapping region is segmented for the preoperative point set. Furthermore, we propose two methods to filter out the non-overlapping regions, at feature-level (i.e., Bi-NOFNet(Feature)) and point-level (i.e., Bi-NOFNet (Point)). For these two methods, we develop supervised registration strategy where the ground-truth overlap mask and displacement vectors are employed, and weakly-supervised registration strategies where only the ground-truth overlap mask is available. Additionally, to fully utilise the information in both space, we propose a bidirectional registration mechanism, which predicts the displacement vectors associated with the intraoperative point set (i.e., the forward way) and those warpping the preoperative point set (i.e., the backward way). Experiments have been conducted on the proposed DeformMedShapeNet dataset that contains 615 different liver shapes. Extensive results demonstrate that Bi-NOFNet performs well for partial-to-full registration tasks under various scenarios of noise, overlap ratios and deformation levels, outperforming existing non-rigid registration approaches.

Three dimensional (3D) overhead crane is a typical multi-input multi-output (MIMO) and underactuated mechatronic system. Complex double spherical pendulum dynamics and varying cable length increase the difficulty of the swing suppression control for overhead crane. Moreover, it would be a great challenge for considering friction uncertain and overshoot issues of the actuators. In this article, an adaptive anti-swing controller is proposed for seven degrees of freedom (7-DOF) overhead crane with double spherical pendulum and varying cable length. First, we use Lagrange’s method to establish an accurate dynamic model of 7-DOF overhead crane. The complex nonlinear dynamic model includes trolley moving, bridge moving, cable length varying, hook swing and payload swing in 3D space. Then, by analysis the system energy function, an adaptive control method is designed to control the trolley, bridge and suspension cable simultaneously. Moreover, we elaborately design some nonlinear terms (such as overshoot limiting term, adaptive term and anti-swing term), which can be used to handle the actuator overshoots, friction uncertain and unactuated swing suppression. As far as we know, it is the first nonlinear closed-loop controller for 7-DOF overhead crane without any linearization. Finally, a group of the hardware experiment results prove that the proposed adaptive anti-swing controller has better effectiveness and robustness than the existing state-of-the-art controllers.Note to Practitioners—This paper studies the anti-swing control problem of the 7-DOF overhead crane system. To improve the efficiency of overhead crane, the trolley moving, the bridge moving and the payload hoisting/lowing are carried out simultaneously in practice. Due to the complex dynamics of the double spherical pendulum and varying cable length, the anti-swing control of the 7-DOF crane remains an open problem. Moreover, most existing control methods ignore the friction uncertain and actuator overshoot, which is not feasible in practice. To tackle these problems, this paper designs a novel adaptive anti-swing controller, which can solve the issues of friction estimation, overshoot limitation, actuators positioning while suppressing the unactuated swing in the 7-DOF overhead crane system. The control performance of the proposed anti-swing controller is verified on the self-build 7-DOF overhead crane platform in the laboratory. In the future, the proposed anti-swing controller will be applied in the industrial overhead crane system to improve the control performance.

Quadruped locomotion is challenging for many learning-based algorithms. This is because it requires tedious manual tuning to cope with different types of terrains and is difficult to deploy in reality due to the sim-to-real gap between the training and the testing scenarios. This paper proposes a quadruped robot learning system for agile locomotion which does not require any pre-training and works well in various terrains. We introduce a hierarchical framework that uses reinforcement learning as the high-level policy to adjust the low-level trajectory generator for a better adaptability to various terrains. We compact the observation and the action spaces of reinforcement learning to deploy the proposed framework on a host computer interfaced with the robot. Besides, we design an omnidirectional trajectory generator guided by robot posture, which generates omnidirectional foot trajectories to interact with the environment. Experimental results and the supplementary video demonstrate that our hierarchical framework only trained in simulation can be easily deployed in the real world, and also has the advantages of fast convergence and good terrain adaptability.Note to Practitioners—This paper presents a hierarchical framework for quadruped robots. It combines a high-level reinforcement learning controller with a posture-guided trajectory generator to adaptively generate omnidirectional motions. Our method is easy to train as it converges fast and does not need to adjust a dozen or so of rewards. The quadruped robot can be deployed in a real environment directly after being trained in simulation. With the trained hierarchical framework deployed on a remote host computer, the robot works well in a variety of real-world environments unseen in the simulation.

Online 3D Bin Packing Problem (3D-BPP) has a wide range of industrial applications and there is an emerging research interest in learning optimal bin packing policy and deploying it for real logistics applications. From the heuristic methods to the deep reinforcement learning (DRL) methods, the previous works have proposed many solutions to solve the online 3D-BPP. However, none of them have studied what and how heuristics can be modelled into DRL to build a more effective and practical bin packing pipeline. In this work, we thoroughly investigate what heuristics can be used in online 3D-BPP and how to effectively integrate the heuristics with the DRL. First, we design 3 different heuristics based on the physical rules of the real world and the experiences of the human packers, including the Physics-Heuristics, the Packing-Heuristics and the Unpacking-Heuristics. Second, we model the 3 types of heuristics into the DRL framework and propose a novel heuristic DRL method to solve the online 3D-BPP. Extensive experimental results show that our method achieves state-of-the-art bin packing performance and the resulting real-world system is able to reliably finish the bin packing task in real logistics scenarios. Supplementary video is available athttps://www.youtube.com/watch?v=x8GpmEELq18. Note to Practitioners—The rapid growth of e-commerce has significantly increased the burden of human packers in logistic warehouses, where the workers need to pick the products from a conveyor and pack them into bins (i.e. the online 3D bin packing). Thus it is of great importance to develop intelligent robotic systems to replace human labor, which is a long-standing topic in the field of control and automation science. This paper makes a substantial contribution to the related field by studying the online 3D bin packing in terms of both the theory and practice. On the one hand, the simulated experiments suggest that the presented algorithm significantly improves the space utilization of bin packing. On the other hand, the robotic system developed based on the proposed method can favourably finish the bin packing task in real logistics scenarios, demonstrating the practical use of our approach. Consequently, the approach proposed in this paper is totally applicable in logistic warehouses and is promising to drastically improve the working efficiency of the product packing in real warehouses. In the future, we will extend the presented approach to pack irregular-shaped objects and then facilitate more logistics applications.

Existing in-pipe robots have insufficient adaptability when dealing with accidents in unfamiliar pipe environments. Developing a pipe robot that can be designed and manufactured quickly is one solution. The tensegrity structure is a self-stressing spatial structure formed by the interaction of rigid members and flexible cables, which has the advantages of simple structure, good flexibility, deformability, and impact resistance. Inspired by this structure, we design a novel worm-like tensegrity robot for different pipe environments, which can be manufactured rapidly at low cost. Firstly, a robotic module based on the tensegrity structure is designed inspired by the motion patterns of worm-like organisms. Then, the design process of the module is presented based on the mathematical analysis of the deformation. Finally, a prototype of the tensegrity robot is developed using simple and low-cost parts in less than an hour. To test the motion performance, load performance, and inspection capability of the tensegrity robot, we designed a series of experiments on horizontal pipes, vertical pipes, elbows, and steel pipes. Experimental results show that the worm-like tensegrity robot is simple in structure, easy to manufacture, low in cost, and good in performance.

We present a real-time, high-accuracy, robust, tightly coupled visual-inertial odometry (VIO) algorithm, including monocular-inertial odometry and stereo-inertial odometry, and uses inertial measurement unit (IMU) pre-integration that is based on fourth-order Runge–Kutta (PK4) and IMU initialization based on maximum a posteriori (MAP) estimation. In particular, we used the multi-state constraint Kalman filter (MSCKF) to fuse vision and IMU measurement data for state estimation. In the optimization stage, we simultaneously considered and optimized all of the historical constraints, and performed multiple iterations to reduce the linearity errors. For further reducing the cumulative error and improving the relocation accuracy, we used a bag-of-words model for global optimization. To lower the computational cost and increase the real-time performance, we set keyframe insertion mechanism and introduced sliding window, and used a new form of Kalman gain that converts the Kalman gain in multi-state constraint Kalman filtering into the inverse of the state dimension. We validated the proposed method by using the EuRoC MAV dataset and KITTI dataset. We performed physics experiments in an outdoor environment with unstable light, to further validate the accuracy and robustness of our method.

Skeleton-based action recognition has been substantially driven by the development of artificial intelligence technology and deep sensors. Recently, graph convolutional networks (GCNs) have achieved excellent performances in skeleton-based action recognition. However, the performances of GCN-based methods are impaired by inappropriate node partitioning strategy and obstructed long-range information flow. To solve these issues, a novel Select-Assemble-Normalize Graph Convolution Network (SAN-GCN) is proposed to model the spatio-temporal features of skeleton. First, all skeleton joints are selected as root nodes, and the neighborhoods of the root joints are assembled and normalized according to the body structure, which explicitly and interpretably expresses the spatial geometric relation of the skeleton joints. Second, we propose an attention-based assembly and normalization strategy to adaptively capture non-local joints. The adaptive assembly and normalization can avoid the dilution of key long-range features. Moreover, a bi-level aggregation strategy is introduced to learn spatio-temporal dependencies of joints, where the low-level aggregation aligns the normalized neighborhood graphs, and the high-level aggregation aggregates the features of neighbor nodes by a standard convolution kernel. In high-level aggregation, it is convenient to realize factorized spatio-temporal aggregation or unified spatio-temporal aggregation. Extensive experiments on four datasets with different numbers of action patterns demonstrate that our model achieves comparable performance with the state-of-the-art works.

In this article, we present the circular accessible depth (CAD), a robust traversability representation for an unmanned ground vehicle (UGV) to learn traversability in various scenarios containing irregular obstacles. To predict CAD, we propose a neural network, namely CADNet, with an attention-based multiframe point cloud fusion module, stability-attention module (SAM), to encode the spatial features from point clouds captured by LiDAR. CAD is designed based on the polar coordinate system and focuses on predicting the border of traversable area. Since it encodes the spatial information of the surrounding environment, which enables a semisupervised learning for the CADNet, and thus, desirably avoids annotating a large amount of data. Extensive experiments demonstrate that CAD outperforms baselines in terms of robustness and precision. We also implement our method on a real UGV and show that it performs well in real-world scenarios.

Learning from demonstration holds the promise of enabling robots to learn diverse actions from expert experience. In contrast to learning from observation-action pairs, humans learn to imitate in a more flexible and efficient manner: learning behaviors by simply “watching.” In this article, we propose a “watch-and-act” imitation learning pipeline that endows a robot with the ability of learning diverse manipulations from visual demonstrations. Specifically, we address this problem by intuitively casting it as two subtasks: 1) understanding the demonstration video and 2) learning the demonstrated manipulations. First, a captioning module based on visual change is presented to understand the demonstration by translating the demonstration video into a command sentence. Then, to execute the captioning command, a manipulation module that learns the demonstrated manipulations is built upon an instance segmentation model and a manipulation affordance prediction model. We validate the superiority of the two modules over existing methods separately via extensive experiments and demonstrate the whole robotic imitation system developed based on the two modules in diverse scenarios using a real robotic arm. Supplementary video is available athttps://vsislab.github.io/watch-and-act/.

This study focuses on localization and mapping for UGVs when they are deployed in environments with non-level roads. In these scenarios, the vehicles need to travel through flat but not necessarily level grounds, i.e., ascent or descent, which may cause drifts of the robot pose and distortion of the map. We develop a low-drift LiDAR odometry and mapping approach for the UGV with LiDAR as the only exteroceptive sensor. A factor-graph based pose optimization method is developed with a specifically designed factor named slope factor. This factor includes the slope information that is estimated from a real-time LiDAR data stream. The slope information is also used to enhance the loop-closure detection procedure. Moreover, an incremental pitch estimation mechanism is designed to achieve further pose estimation refinement. We demonstrate the effectiveness of the developed framework in real-world environments. The odometry drift is lower and the map is more precise than experiments with the state-of-the-arts. Notably, on the Kitti dataset, our method also exhibits convincing performance, demonstrating its strength in more general application scenarios.

This paper presents a novel concept to develop robots capable of crawling in tubular environments, inspired by the movement of earthworms and the biological musculoskeletal systems in nature. A tensegrity structures-based robotic module with shape changeability actuated by only one linear actuator is proposed. The mechanical structure of the robotic module is determined on the basis of force density method. By serially cascading three uniform modules, the in-pipe crawling robot is designed and manufactured. The robot has the abilities to crawl in both horizontal and vertical pipes with different inner diameters, and to pass through elbow pipes adaptively under the control of a simple actuation sequence. The effectiveness of the robot is demonstrated by experimental results on the prototype. Compared with existing robots, this proposed approach enables compact yet robust structures, along with enhanced compliance, mobility, and adaptability.

Extracting effective spatial-temporal information is significantly important for video-based action recognition. Recently 3D convolutional neural networks (3D CNNs) that could simultaneously encode spatial and temporal dynamics in videos have made considerable progress in action recognition. However, almost all existing 3D CNN-based methods recognize human actions only using RGB videos. The single modality may limit the performance capacity of 3D networks. In this paper, we extend 3D CNN to depth and pose data besides RGB data to evaluate its capacity for spatiotemporal multimodal learning for video action recognition. We propose a novel multimodal two-stream 3D network framework, which can exploit complementary multimodal information to improve the recognition performance. Specifically, we first construct two discriminative video representations under depth and pose data modalities respectively, referred as depth residual dynamic image sequence (DRDIS) and pose estimation map sequence (PEMS). DRDIS captures spatial-temporal evolution of actions in depth videos by progressively aggregating the local motion information. PEMS eliminates the interference of cluttered backgrounds and describes the spatial configuration of body parts intuitively. The multimodal two-stream 3D CNN deals with two separate data streams to learn spatiotemporal features from DRDIS and PEMS representations. Finally, the classification scores from two streams are fused for action recognition. We conduct extensive experiments on four challenging action recognition datasets. The experimental results verify the effectiveness and superiority of our proposed method.

A method for planning and controlling the somersault motion of a quadruped robot is proposed in this paper. The method divides the somersault motion into 5 stages according to intuitive understanding. Based on the simplified dynamic model, the linear programming method is used to obtain the maximum ground reaction force under the constraints of joint torque and friction cone, and then the optimal leg thrusting trajectory is obtained by double integration of the acceleration. In order to achieve the buffered landing of the robot after somersault, a whole body controller based on null space projection is used to obtain the optimal joint torque under the constraints of the robot’s foot position, torso position and torso posture. The somersault motion control method proposed in this paper has been verified by the dynamics simulation software Webots and quadruped robot platform Yobogo. The results show that the robot can complete stable front flip and back flip under the constraints of joint output torque and foot motion space constraints.

In this paper, we propose a novel deep reinforcement learning (DRL) system for the autonomous navigation of mobile robots that consists of three modules: map navigation, multi-view perception and multi-branch control. Our DRL system takes as the input a routed map provided by a global planner and three RGB images captured by a multi-camera setup to gather global and local information, respectively. In particular, we present a multi-view perception module based on an attention mechanism to filter out redundant information caused by multi-camera sensing. We also replace raw RGB images with low-dimensional representations via a specifically designed network, which benefits a more robust sim2real transfer learning. Extensive experiments in both simulated and real-world scenarios demonstrate that our system outperforms state-of-the-art approaches.

This paper focuses on developing a real-time and flexible velocity estimation approach for serial revolute manipulator using only one inertial measurement unit (IMU) mounted on each link side of the manipulator. Particularly, the proposed approach has no requirement for the installation position and orientation of the IMU, which improves the flexibility of the implementation procedure. A joint velocity model is established based on the proposed principle of constructing coordinate system according to the robotic geometric information. The general solutions are derived in detail, thereby the proposed algorithm can be generalized into any other robots with the same geometric configuration. With the method, the joint rotational velocity measurements of static and dynamic robotic motion are provided compared to encoders. Experimental results based on the six degrees of freedom (DOF) collaborative manipulator have validated the feasibility and effectiveness of the proposed approach. The proposed method has the benefits of low cost and flexibility, which could work as a redundant velocity monitor criterion to provide assistant joint velocity measurements.

The bones, muscles, tendons and connective tissues form a continuous tension network throughout human body. This heterogeneous mixture presents the characteristics of tensegrity, providing the body with structurally integrity, stability and flexibility. Inspired by this, this paper proposes a novel soft robotic gripper based on tensegrity structures. Firstly, the design and working principle of the tensegrity-based robotic gripper is introduced, which is composed of a series of discrete rigid segments connected with tensegrity joints by means of tensional cables. Then, the kinematics of the robotic gripper is analyzed using force density method to obtain the relationship between the pose of the gripper and the tension of cables. Finally experiments on the developed prototype demonstrates that the robotic gripper is able to grasp various objects of different sizes, shapes, and materials. Additional desirable properties are derived from using tensegrity structures in the robotic gripper: light weight, high compliance, inherent safety, low cost, and waterproof and dustproof performance. It is suggested that tensegrity structures have great potential to be an effective alternative to the development of soft robotic grippers.

Quadruped locomotion is a challenging task for learning-based algorithms. It requires tedious manual tuning and is difficult to deploy in reality due to the reality gap. In this paper, we propose a quadruped robot learning system for agile locomotion which does not require any pre-training and works well in various real-world terrains. We introduce a hierarchical learning framework that uses reinforcement learning as the high-level policy to adjust the low-level trajectory generator for better adaptability to the terrain. We compact the observation and action space of the reinforcement learning to deploy it on a host computer in reality. Besides, we design a trajectory generator guided by robot posture, which can generate adaptive foot trajectory to interact with the environment. Experimental results show that our system can be easily deployed in reality while only trained in simulation, and also has the advantages of fast convergence and good terrain adaptability. The supplementary video demonstration is available at https://vsislab.github.io/hfql/.

Product packing is a typical application in ware-house automation that aims to pick objects from unstructured piles and place them into bins with optimized placing policy. However, it still remains a significant challenge to finish the product packing tasks in general logistics scenarios where the objects are variable-sized and the configurations are complex. In this work, we present the PackerBot, a complete robotic pipeline for performing variable-sized product packing in unstructured scenes. First, by leveraging the imperfect experience of human packer, we propose a heuristic DRL framework for learning optimal online 3D bin packing policy. Then we integrate it with a 6-DoF suction-based picking module and a product size estimation module, leading to a complete product packing system, namely the PackerBot. Extensive experimental results show that our method achieves the state-of-the-art performance in both simulated and real-world tests. The video demonstration is available at: https://vsislab.github.io/packerbot.

Compared with torque-control techniques, a position-controlled quadruped robot is lower cost, easier to build, and more direct to drive. However, the stiff actuation of position-controlled actuators makes it difficult for the quadruped to achieve dynamically stable locomotion. This paper presents an implementation of joint velocity programming technique to regulate the body’s moving speed and orientation for a position-controlled quadruped robot that performs trotting or pacing locomotion. The robot model is mapped to a new coordinate space in order to decouple the control of its body. In one plane of the new coordinate space, the robot is simplified to an inverted pendulum model to generate attitude and velocity tracking actions. In the other planes, body regulating problems are formulated in velocity forms and solved by designing support leg motions. The controllers in these planes are integrated to produce joint velocities that enable robust trotting and pacing locomotion at a variety of speeds and directions, despite lacking force control or feedback techniques. Physical test results as well as simulating results demonstrate control of the quadruped robot SmarQ to perform omni-directional locomotion, impact recovery, and adaptability to uneven terrains.

The task of spatial layout estimation of monocular image is to segment an RGB image of indoor scenes with semantic surface labels (i.e., ceiling, floor, front wall, left wall, and right wall). Most recent methods have to produce layout hypotheses based on the estimated edge map or semantic labels, and then rank the layout hypotheses. In this paper, we present an end-to-end framework that can directly output the layout type and keypoint coordinates (defined in the LSUN challenge). The proposed method takes advantage of transfer learning via learning on the fake samples, i.e., plenty of artificial {type, keypoints, edge map} triplets are generated to learn the mapping from edge maps to keypoint coordinates. Generative adversarial network (GAN) is implemented in this work for domain adaptation of the edge maps. Experimental results show that the proposed method can achieve state-of-the-art layout estimation performance on benchmark datasets.

Imitation learning enables robots to learn a task by simply watching the demonstration of the task. Current imitation learning methods usually require the learner and demonstrator to occur in the same context. This limits their scalability to practical applications. In this paper, we propose a more general imitation learning method which allows the learner and the demonstrator to come from different contexts, such as different viewpoints, backgrounds, and object positions and appearances. Specifically, we design a robotic system consisting of three models: context translation model, depth prediction model and multi-modal inverse dynamics model. First, the context translation model translates the demonstration to the context of learner from a different context. Then combining the color observation and depth observation as inputs, the inverse model maps the multi-modal observations into actions to reproduce the demonstration, where the depth observation is provided by a depth prediction model. By performing the block stacking tasks both in simulation and real world, we prove the cross-context learning advantage of the proposed robotic system over other systems.

Integrated robotic arm system should contain both grasp and place actions. However, most grasping methods focus more on how to grasp objects, while ignoring the placement of the grasped objects, which limits their applications in various industrial environments. In this research, we propose a model-free deep Q-learning method to learn the grasping-stacking strategy end-to-end from scratch. Our method maps the images to the actions of the robotic arm through two deep networks: the grasping network (GNet) using the observation of the desk and the pile to infer the gripper's position and orientation for grasping, and the stacking network (SNet) using the observation of the platform to infer the optimal location when placing the grasped object. To make a long-range planning, the two observations are integrated in the grasping for stacking network (GSN). We evaluate the proposed GSN on a grasping-stacking task in both simulated and real-world scenarios.

We present an imitation learning method for autonomous drone patrolling based only on raw videos. Different from previous methods, we propose to let the drone learn patrolling in the air by observing and imitating how a human navigator does it on the ground. The observation process enables the automatic collection and annotation of data using inter-frame geometric consistency, resulting in less manual effort and high accuracy. Then a newly designed neural network is trained based on the annotated data to predict appropriate directions and translations for the drone to patrol in a lane-keeping manner as humans. Our method allows the drone to fly at a high altitude with a broad view and low risk. It can also detect all accessible directions at crossroads and further carry out the integration of available user instructions and autonomous patrolling control commands. Extensive experiments are conducted to demonstrate the accuracy of the proposed imitating learning process as well as the reliability of the holistic system for autonomous drone navigation. The codes, datasets as well as video demonstrations are available at https://vsislab.github.io/uavpatrol.

A floating base system is inevitably to contact the environment while it is moving. This paper explores the contact force estimation and regulation algorithm for a position-controlled floating base system without joint torque information. First, the joint space dynamic model of the system is presented and transformed into the contact space. Then, the inverse dynamics method is employed to estimate the contact forces. After that, a proportional-integral (PI) regulator is designed to drive the contact forces to track the desired values. Finally, the feasibility of this algorithm is demonstrated on a simulated bipedal platform.

In this paper, we propose an autonomous method for robot navigation based on a multi-camera setup that takes advantage of a wide field of view. A new multi-task network is designed for handling the visual information supplied by the left, central and right cameras to find the passable area, detect the intersection and infer the steering. Based on the outputs of the network, three navigation indicators are generated and then combined with the high-level control commands extracted by the proposed MapNet, which are finally fed into the driving controller. The indicators are also used through the controller for adjusting the driving velocity, which assists the robot to adjust the speed for smoothly bypassing obstacles. Experiments in real-world environments demonstrate that our method performs well in both local obstacle avoidance and global goal-directed navigation tasks.

A deep visual tracker is typically based on either object detection or template matching while each of them is only suitable for a particular group of scenes. It is straightforward to consider fusing them together to pursue more reliable tracking. However, this is not wise as they follow different tracking principles. Unlike previous fusion-based methods, we propose a novel ensemble framework, named DTNet, with an online decision mechanism for visual tracking based on hierarchical reinforcement learning. The decision mechanism substantiates an intelligent switching strategy where the detection and the template trackers have to compete with each other to conduct tracking within different scenes that they are adept in. Besides, we present a novel detection tracker which avoids the common issue of incorrect proposal. Extensive results show that our DTNet achieves state-of-the-art tracking performance as well as good balance between accuracy and efficiency. The project website is available at https://vsislab.github.io/DTNet/.

Visual object tracking is a fundamental and time-critical vision task. However, most trackers such as SiamFC and CFNet missed the object movement and simply defined the searching region centered at the location of the target in the previous frame. So they tend to fail in the cases with severe occlusion or a large displacement of the target. In this paper, we consider the object tracking as a dual-task problem of guessing and matching. A guess module is to estimate the motion trend of the target by reinforcement learning based on the observations on appearance changes and motion history. Rather than using the previous location of the target, we may have a more accurate center to locate the searching region. Benefited from such improved searching region, the match module becomes less prone to the object drift problem, and can easily identify the target from the potential distractors in the background. Extensive experimental results on benchmark datasets such as RGBT, OTB-2013, OTB-50 and OTB-100, show that the proposed method achieves leading performance compared to state-of-the-art trackers. Moreover, the proposed tracker could maintain real-time speed, giving itself the potential in practical applications.

With the help of convolutional neural networks (CNNs), video-based human action recognition has made significant progress. CNN features that are spatial and channel-wise can provide rich information for powerful image description. However, CNNs lack the ability to process the long-term temporal dependency of an entire video and further cannot well focus on the informative motion regions of actions. Aiming at the two problems, we propose a novel video-based action recognition framework in this paper. We first represent videos with dynamic image sequences (DISs), which effectively describe videos by modeling the local spatial-temporal dynamics and dependencies. Then a channel and spatial-temporal interest points (STIPs) attention model (CSAM) based on CNNs is proposed to focus on the discriminative channels in networks and the informative spatial motion regions of human actions. Specifically, channel attention (CA) is implemented by automatically learning channel-wise convolutional features and assigning different weights for different channels. STIPs attention (SA) is encoded by projecting the detected STIPs on frames of dynamic image sequences into the corresponding convolutional feature map space. The proposed CSAM is embedded after CNN convolutional layers to refine the feature maps, followed by global average pooling to produce effective feature representations for videos. Finally frame-level video representations are fed into an LSTM to capture the temporal dependencies and make classification. Experiments on three challenging RGB-D datasets show that our method has better performance and outperforms the state-of-the-art approaches with only depth data.

Nonprehensile manipulation produces underconstraint motions that are sensitive to environmental dynamics. Legged locomotion constitutes a floating-based movement, whose dynamic is underactuated with respect to the inertial frame. When these two tasks are combined, system motion planning and control are complex due to their inherent underactuated features. This article presents a motion planning framework for a legged robot that uses its limbs for nonprehensile manipulation, as well as locomoting motions. First, issues related to the description of the robot-object-environment system and the task are presented. The velocity constraint that prevents separation and the force constraint that restricts interactive forces are then integrated into the system dynamic model to produce bounds on the system acceleration as a function of the system state. Then, we solve the motion planning problem by reducing the system dimensions in operational space and programming feasible trajectories within the phase plane. This approach is employed to control the quadruped robot TITAN-VIII to manipulate objects and locomote itself using Drive Mode, Inchworm Mode, Scoot Mode, and Throw Mode. Experimental results obtained through simulations and physical tests are reported to demonstrate the effectiveness of our approach.

Learning actions from human demonstration is an emerging trend for designing intelligent robotic systems, which can be referred as video to command. The performance of such approach highly relies on the quality of video captioning. However, the general video captioning methods focus more on the understanding of the full frame, lacking of consideration on the specific object of interests in robotic manipulations. We propose a novel deep model to learn actions from human demonstration video for robotic manipulation. It consists of two deep networks, grasp detection network (GNet) and video captioning network (CNet). GNet performs two functions: providing grasp solutions and extracting the local features for the object of interests in robotic manipulation. CNet outputs the captioning results by fusing the features of both full frames and local objects. Experimental results on UR5 robotic arm show that our method could produce more accurate command from video demonstration than state-of-the-art work, thereby leading to more robust grasping performance.

The generative adversarial network has been shown to produce state‐of‐the‐art results of image generation. In this study, the authors propose a novel adversarial training method to train salient object detection (SOD) models. They train a convolutional SOD network along with a gated adversarial network that discriminates salient maps coming either from the ground truth or from the SOD network. The motivation for our approach is that the adversarial network can detect and correct pixel‐wise errors between ground truth salient detection maps and the ones produced by the convolutional network. Our experiments show that the adversarial training approach leads to state‐of‐the‐art performance on MSRA‐B, extended complex scene saliency dataset, HKU‐IS, DUT, and SOD dataset.

The aspect ratio of a target changes frequently during an unmanned aerial vehicle (UAV) tracking task, which makes the aerial tracking very challenging. Traditional trackers struggle from such a problem as they mainly focus on the scale variation issue by maintaining a certain aspect ratio. In this paper, we propose a coarse-to-fine deep scheme to address the aspect ratio variation in UAV tracking. The coarse-tracker first produces an initial estimate for the target object, then a sequence of actions are learned to fine-tune the four boundaries of the bounding box. The coarse-tracker and the fine-tracker are designed to have different action spaces and operating target. The former dominates the entire bounding box and the latter focuses on the refinement of each boundary. They are trained jointly by sharing the perception network with an end-to-end reinforcement learning architecture. Experimental results on benchmark aerial data set prove that the proposed approach outperforms existing trackers and produces significant accuracy gains in dealing with the aspect ratio variation in UAV tracking.

Video-based person re-identification (re-id) matches two tracks of persons from different cameras. Features are extracted from the images of a sequence and then aggregated as a track feature. Compared to existing works that aggregate frame features by simply averaging them or using temporal models such as recurrent neural networks, we propose an intelligent feature aggregate method based on reinforcement learning. Specifically, we train an agent to determine which frames in the sequence should be abandoned in the aggregation, which can be treated as a decision making process. By this way, the proposed method avoids introducing noisy information of the sequence and retains these valuable frames when generating a track feature. On benchmark data sets, experimental results show that our method can boost the re-id accuracy obviously based on the state-of-the-art models.

This paper presents the dynamic modelling and motion planning method for a quadruped robot that uses its legs for nonprehensile manipulation as well as locomotion. Three different working modes named Drive Mode, Inchworm Mode and Scoot Mode are proposed to enable the robot to move forward together with the object. We firstly introduce a universal model for these modes and deduce its dynamic equation. Then the contact force constraints are combined and mapped to the system state variables. Based on the acquired state acceleration constraints, the motion planning problem can be solved by designing system state paths in the phase space. After that, we described the mathematical problems within the three working modes and generate the robot motions accordingly. Finally, experimental results obtained through simulations and physical tests are reported to demonstrate the effectiveness of our method.

This paper explores the control strategy for a multi-limb robot nonprehensilely pushing an object to slide on the floor. The robot's limb distals perform point contacts with the object and the floor. The contact velocity constraint and force constraint are proposed to prevent separation and restrict the system forces. Then the constraints are combined with the system dynamic models to obtain bounds on the system states. We solve the motion planning problem by selecting a feasible path in the reduced-dimensional space and generating the system trajectory along the selected path. An example is provided to illustrate the application of our technique on the physical platform.

A novel gait planning and control framework was developed for quadruped locomotion of a robot. It modeled the quadruped robot as two bipeds carrying the body from the front and rear ends. We first mapped the relationship between the joint torques of support legs and the torso forces of the bipedal sub-robots. Then the equations describing the relationship between the quadruped body forces and the bipedal torso forces under various operating modes of the robot were deduced and solved. Virtual forces were generated on the quadruped body to manipulate its velocity and orientation. Then these virtual forces were distributed to the front and hind sub-robots to generate support leg torques. The state machines and gait generators for the two bipedal sub-robots were designed individually, resulting in the decoupling of the gait parameters in the front legs and hind legs. The effectiveness of the controller was validated through dynamic simulations.

Object tracking is an important capability for robots tasked with interacting with humans and the environment, and it enables robots to manipulate objects. In object tracking, selecting samples to learn a robust and efficient appearance model is a challenging task. Model learning determines both the strategy and frequency of model updating, which concerns many details that can affect the tracking results. In this paper, we propose an object tracking approach by formulating a new objective function that integrates the learning paradigm of self-paced learning into object tracking such that reliable samples can be automatically selected for model learning. Sample weights and model parameters can be learned by minimizing this single objective function under the framework of kernelized correlation filters. Moreover, a real-valued error-tolerant self-paced function with a constraint vector is proposed to combine prior knowledge, i.e., the characteristics of object tracking, with information learned during tracking. We demonstrate the robustness and efficiency of our object tracking approach on a recent object tracking benchmark data set: OTB 2013.

We introduce a novel strategy of designing a chaotic coverage path planner for the mobile robot based on the Chebyshev map for achieving special missions. The designed chaotic path planner consists of a two-dimensional Chebyshev map which is constructed by two one-dimensional Chebyshev maps. The performance of the time sequences which are generated by the planner is improved by arcsine transformation to enhance the chaotic characteristics and uniform distribution. Then the coverage rate and randomness for achieving the special missions of the robot are enhanced. The chaotic Chebyshev system is mapped into the feasible region of the robot workplace by affine transformation. Then a universal algorithm of coverage path planning is designed for environments with obstacles. Simulation results show that the constructed chaotic path planner can avoid detection of the obstacles and the workplace boundaries, and runs safely in the feasible areas. The designed strategy is able to satisfy the requirements of randomness, coverage, and high efficiency for special missions.

The trajectories of mobile seeds have a great influence on localization accuracy and efficiency. This paper presents a novel information-driven intelligent mobility-assisted wireless sensor network localization algorithm. Without requiring any prior knowledge of the sensing field, seeds' or pseudo-seeds' (common sensors which have been positioned) trajectories are scheduled dynamically aiming at position estimates of neighboring non-positioned common sensors. With an information-theoretic utility measure as the objective function, mobile seeds or pseudo-seeds actively determine their motion directions for minimizing the uncertainty in position estimates of neighboring sensors. At the first level, seeds estimate the neighboring sensor nodes' positions with bearing measurements by means of extended Kalman filters and optimize their motion directions by maximizing the mutual information between the position estimates and the motions of seeds. Afterwards the seeds forward the position estimates to the corresponding sensor nodes, which then act as pseudo-seeds. By repeating this process at the following levels, all sensor nodes can obtain position estimates. Compared with heuristic mobility and random mobility-assisted mobile sensor network localization algorithms, the proposed algorithm requires fewer maneuvers of seed or pseudo-seeds for quick convergence to good position estimates. Extensive simulations show that this algorithm can provide more accurate position estimates with fewer maneuvers, especially in the case of limited seeds.

Falls are one of the major causes leading to injury of elderly people. Using wearable devices for fall detection has a high cost and may cause inconvenience to the daily lives of the elderly. In this paper, we present an automated fall detection approach that requires only a low-cost depth camera. Our approach combines two computer vision techniques-shape-based fall characterization and a learning-based classifier to distinguish falls from other daily actions. Given a fall video clip, we extract curvature scale space (CSS) features of human silhouettes at each frame and represent the action by a bag of CSS words (BoCSS). Then, we utilize the extreme learning machine (ELM) classifier to identify the BoCSS representation of a fall from those of other actions. In order to eliminate the sensitivity of ELM to its hyperparameters, we present a variable-length particle swarm optimization algorithm to optimize the number of hidden neurons, corresponding input weights, and biases of ELM. Using a low-cost Kinect depth camera, we build an action dataset that consists of six types of actions (falling, bending, sitting, squatting, walking, and lying) from ten subjects. Experimenting with the dataset shows that our approach can achieve up to 91.15% sensitivity, 77.14% specificity, and 86.83% accuracy. On a public dataset, our approach performs comparably to state-of-the-art fall detection methods that need multiple cameras.

This paper deals with a new approach based on Q -learning for solving the problem of mobile robot path planning in complex unknown static environments. As a computational approach to learning through interaction with the environment, reinforcement learning algorithms have been widely used for intelligent robot control, especially in the field of autonomous mobile robots. However, the learning process is slow and cumbersome. For practical applications, rapid rates of convergence are required. Aiming at the problem of slow convergence and long learning time for Q -learning based mobile robot path planning, a state-chain sequential feedback Q -learning algorithm is proposed for quickly searching for the optimal path of mobile robots in complex unknown static environments. The state chain is built during the searching process. After one action is chosen and the reward is received, the Q -values of the state-action pairs on the previously built state chain are sequentially updated with one-step Q -learning. With the increasing number of Q -values updated after one action, the number of actual steps for convergence decreases and thus, the learning time decreases, where a step is a state transition. Extensive simulations validate the efficiency of the newly proposed approach for mobile robot path planning in complex environments. The results show that the new approach has a high convergence speed and that the robot can find the collision-free optimal path in complex unknown static environments with much shorter time, compared with the one-step Q -learning algorithm and the Q ( λ )-learning algorithm.

We describe our research in using environmental visual landmarks as the basis for completing simple robot construction tasks. Inspired by honeybee visual navigation behavior, a visual template mechanism is proposed in which a natural landmark serves as a visual reference or template for distance determination as well as for navigation during collective construction. To validate our proposed mechanism, a wall construction problem is investigated and a minimalist solution is given. Experimental results show that, using the mechanism of a visual template, a collective robotic system can successfully build the desired structure in a decentralized fashion using only local sensing and no direct communication. In addition, a particular variable, which defines tolerance for alignment of the structure, is found to impact the system performance. By decreasing the value of the variable, system performance is improved at the expense of a longer construction time. The visual template mechanism is appealing in that it can use a reference point or salient object in a natural environment that is new or unexplored and it could be adapted to facilitate more complicated building tasks.