Intelligent Service Robotics最新文献

The indoor positioning problem is a critical research domain essential for real-time control of mobile robots. Within this field, Monte Carlo-based solutions have been devised, leveraging the processing of diverse sensor data to address numerous challenges in local and global positioning. This study focuses on resampling strategies within the conventional Monte Carlo framework, which directly impact positioning performance. From this perspective, in contrast to the conventional method of employing weight thresholding and full particle scattering when the robot becomes disoriented, this study proposes an alternative approach. It advocates for a localized space resampling strategy, adaptive noise injection guided by likelihood, and the incorporation of beam rejection modifications to address dynamic (unmapped) obstacles effectively. The real-time experimental results, conducted with varying particle counts, demonstrate that the proposed scheme effectively manages the presence of unmapped obstacles while employing fewer particles than the standard Monte Carlo implementation.

Abstract

In today’s world, the utilization of a large number of vehicles has led to congested traffic conditions and an increase in accidents. These issues are considered primary problems in the transportation field. Therefore, there is a pressing need to develop a novel method for monitoring traffic. To address this, we propose a new model called the residual faster recurrent convolutional (RFRC) algorithm. While the proposed model achieves good detection accuracy, it must also meet the demands of real-life scenarios. In this approach, the ResNet-50 model is combined with the faster recurrent-based convolutional neural network (FRCNN) to enable the detection of autonomous vehicles. We utilize the dung beetle optimizer (DBO) with a crossover strategy for feature selection, focusing on selecting relevant features for analysis. To validate the effectiveness of the proposed RFRC method, we conduct experiments using two datasets: the KITTI dataset and the COCO2017 dataset. The evaluation of the RFRC model is performed using various measures, including f1-score, precision, recall, accuracy, and specificity, on both datasets. The proposed RFRC model outperforms both datasets and attains better results in autonomous vehicle detection.

As AI-enabled robots enter the realm of healthcare and caregiving, it is important to consider how they will address the dimensions of care and how they will interact not just with the direct receivers of assistance, but also with those who provide it (e.g., caregivers, healthcare providers, etc.). Caregiving in its best form addresses challenges in a multitude of dimensions of a person’s life: from physical to social-emotional and sometimes even existential dimensions (such as issues surrounding life and death). In this study, we use semi-structured qualitative interviews administered to healthcare professionals with multidisciplinary backgrounds (physicians, public health professionals, social workers, and chaplains) to understand their expectations regarding the possible roles robots may play in the healthcare ecosystem in the future. We found that participants drew inspiration in their mental models of robots from both works of science fiction but also from existing commercial robots. Participants envisioned roles for robots in the full spectrum of care, from physical to social-emotional and even existential-spiritual dimensions, but also pointed out numerous limitations that robots have in being able to provide comprehensive humanistic care. While no dimension of care was deemed as exclusively the realm of humans, participants stressed the importance of caregiving humans as the primary providers of comprehensive care, with robots assisting with more narrowly focused tasks. Throughout the paper, we point out the encouraging confluence of ideas between the expectations of healthcare providers and research trends in the human–robot interaction (HRI) literature.

Object detection is a primary means of unmanned aerial vehicle (UAV) maritime search and rescue. The problem of scale variation caused by UAV flight height changes, shooting angle changes, and giant waves seriously affects the detection performance. However, most work does not explicitly consider the effects of these factors. In this work, we propose an algorithm called Preprocessing and Attention Scaling, which explicitly considers the scale variation problem caused by height, angle changes, and giant waves for the first time and solves it through Preprocessing Scaling and Attention Scaling. The Preprocessing Scaling module scales and perspective changes the images according to each photograph’s recorded flight altitude and shooting angle and crops them to the appropriate size, significantly improving the detection accuracy and shortening the inference time. At the same time, the scale variation caused by the up and down of the object due to the vast swells cannot be solved by the Preprocessing Scaling module, so we designed the Attention Scaling module again to quickly capture the area that needs further scale change by fusing the horizontal attention and vertical attention, and then transform it to the appropriate scale by the affine transformation, further improving detection accuracy. We extensively tested PAS on the well-known SeaDronesSee-DET and the SeaDronesSee-DET v2 (S-ODv2) datasets, significantly improving the detection accuracy. In addition, we successfully tested our method on a height-angle transfer task, where we trained on some height-angle intervals and tested on different height-angle intervals, achieving good results.

The development and control of a rotary series elastic actuator (SEA) are investigated in this paper. First, a rotary SEA is designed with a small volume and is lightweight, where the elastic element can be used as a torque sensor. We improve the structure of this rubber material elastic element, and its characteristics are analyzed. To provide a more comprehensive description of the entire system, motor dynamics are also taken into account while establishing the entire system dynamics model. Second, a neural network-driven model predictive control (NNMPC) method is proposed for the single-link SEA system. Since a real dynamic system for the SEA is hard to establish accurately due to disturbances, uncertainties, and varying mass of the load in different applications, a simple nonlinear autoregressive neural network using the rectified linear unit as the activation function (ReLU-NARX NN) is considered to approximate the system dynamic model, based on which a model predictive controller is developed. Finally, both numerical simulations and experiments are conducted for position and torque control. The simulation and experimental results demonstrate that the proposed method is superior to the conventional PD (proportional differential) method and the traditional MPC method. For position control, the NNMPC method is shown to be more effective, that is, it can suppress residual vibrations, reduce overshoots, arrive at a steady state quickly, and robust to different loads in a range. For torque control, the control performance is also satisfactory.

Due to random sampling and the unpredictability of moving obstacles, it remains challenging for mobile robots to effectively learn navigation policies and accomplish obstacle avoidance safely. Overcoming such challenges can reduce the time cost required for navigation model training and validation, improving the safety and credibility of autonomous navigation in medical service and industrial patrol. This article proposes an improved soft actor–critic model to enhance the autonomous navigation performance of robots. We first introduce a prioritized experience replay method to reduce the randomness of sampling. The performance of the navigation policy can be enhanced by prioritizing the learning of high-value experiences. Moreover, we also design a network with long short-term memory abilities to store historical environmental information. In this way, temporal characteristics of obstacle motion can be obtained to optimize obstacle avoidance policy. Experimental results in simulation and real-world show that the proposed model significantly improves learning speed, success rate, and trajectory smoothness while exhibiting excellent obstacle avoidance performance in dynamic environments.

In care and nursing education systems, students lack opportunities for acquiring the necessary skills and experiences from practice with actual patients. In this paper, we present a patient robot with musculoskeletal symptoms that supports efficient care education for caregivers to investigate the effects of repetitive range of motion (ROM) exercises. Four students and four experts (who have had many years of experience in the medical field) participated in the data acquisition process by performing repetitive ROM tasks using a patient robot. Based on the collected data, the results were analyzed and the effectiveness and feasibility of repetitive ROM exercises conducted using the patient robot were discussed. This study may provide a new pathway for developing advanced patient robots for use in care training environments by imitating the symptoms of various muscle and joint diseases such as palsy, contractures, and muscle weakness.

Gardening pruning robots are widely applied in green space construction. However, increase of green space environment complexity and obstacle number affect the coverage range and work efficiency of robots. To solve this problem, this research proposed a full-coverage path planning algorithm integrating hybrid genetic ant colony and A* algorithm. Specifically tailored to the lawn working environments of horticultural pruning robots, we initially employed visual simultaneous localization and mapping to create a 3D point cloud map, converting it into an occupancy grid map for future path planning. The obtained grid map was partitioned into multiple subareas on the basis of the locations of obstacles. The optimal traversal order of sub-regions was determined using hybrid genetic ant colony method and a new update strategy of heuristic and pheromone factors was developed for improving the ability of global search and probability of jumping out of local optimal solution. Boustrophedon method was applied to fully cover each sub-region, A* algorithm was adopted to connect various sub-regions, and connection strategy was optimized. Simulation results showed that compared with traditional ant colony algorithm and other full-coverage planning algorithms, the algorithm developed in this research presented superior performance in terms of traversal path length, starting distance, coverage rate and turning times on maps with various sizes and complexities.

Seasonal flu is currently a major public health issue the world is facing. Although the World Health Organization (WHO) suggests social distancing is one of the best ways to stop the spread of the flu disease, the lack of controllability in keeping a social distance is widespread. Spurred by this concern, this paper developed a fast social distancing monitoring solution, which combines a lightweight PyTorch-based monocular vision detection model with inverse perspective mapping (IPM) technology, enabling the nursing robot to recover 3D indoor information from a monocular image and detect the distance between pedestrians, then conducts a live and dynamic infection risk assessment by statistically analyzing the distance between the people within a scene and ranking public places into different risk levels, called Fast DeepSOCIAL (FDS). First, the FDS model generates the probability of an object’s category and location directly using a lightweight PyTorch-based one-stage detector, which enables a nursing robot to obtain significant real-time performance gains while reducing memory consumption. Additionally, the FDS model utilizes an improved spatial pyramid pooling strategy, which introduces more branches and parallel pooling with different kernel sizes, which will be beneficial in capturing the contextual information at multiple scales and thus improving detection accuracy. Finally, the nursing robot introduces a gap-seeking strategy based on obstacles-weighted control (GSOWC) to adapt to dangerous indoor disinfection tasks while quickly avoiding obstacles in an unknown and cluttered environment. The performance of the FDS on the nursing robot platform is verified through extensive evaluation, demonstrating its superior performance compared to seven state-of-the-art methods and revealing that the FDS model can better detect social distance. Overall, a nursing robot employing the Fast DeepSOCIAL model (FDS) will be an innovative approach that effectively contributes to dealing with this seasonal flu disaster due to its fast, contactless, and inexpensive features.

Abstract

Probabilistic roadmap (PRM) method has been shown to perform well in robot path planning. However, its performance degrades when the robot needs to pass through narrow passages. To solve this problem, an improved PRM method with hybrid uniform sampling and Gaussian sampling is proposed in this paper. With the proposed method, the robot can improve the success rate and efficiency of path planning in narrow passages. Firstly, the narrow-passage-aware Gaussian sampling method is developed for narrow passages. Combining uniform sampling globally, the new sampling strategy can increase the sampling density at the narrow passages and reduce the redundancy of the samples in the wide-open regions. Then, we propose to use density-based clustering method to achieve accurate identification of narrow channels by removing the noise points. Next, graph search algorithm is used to search the shortest path from the start point to the goal point. Finally, simulations are carried out to evaluate the validity of the proposed method. Results show that the improved PRM method is more effective for path planning with narrow passages.