Intelligent Service Robotics最新文献

Injuries and diseases such as wrist nerve injuries, stroke, neurological disorders, and other wrist-related conditions have significantly impacted people’s quality of life. This study aims to develop a lightweight, affordable, and portable ForeWrist (forearm and wrist) exoskeleton. This device is intended to assist and rehabilitate individuals with wrist disabilities, mainly stroke survivors, to enhance wrist range of motion and strength. The device can offer one active degree of freedom (DOF) responsible for pronation-supination (PS) of the forearm and two passive DOFs for the wrist joint. The design of the ForeWrist PS mainly consists of a cable-driven C-shaped guide rail and stationary bearing-carriage mechanism that can be attached to the user’s wrist. The simulation and experimental analysis are conducted for the design validation and performance analysis. The experimental results indicate that the designed device should demonstrate promising potential for practical applications. The root mean squared error for joint position and velocity exhibit low values, and the peak torque for an average weight of the human lower arm was found to be under 10% of the device’s total capacity. The developed exoskeleton provides a full range of motion for daily activities and covers 75% of the forearm’s total range of motion with a consistency error of less than (1^circ ). The device can be effective for both at home and outdoor assistance and rehabilitation training with its low weight of 300 g and peak velocity and torque of 70 deg/sec and 6 Nm, respectively.

This paper presents a novel vision-based human–machine interface (HMI) incorporated into an exoskeleton glove tailored for patients with brachial plexus injuries. Addressing the challenges posed by the loss of hand muscle control in individuals affected by these injuries, a fully automated exoskeleton glove function akin to a robotic gripper is used to prevent muscle atrophy through targeted hand muscle exercises. The proposed vision-based HMI is designed for a fully automated exoskeleton glove and incorporates computer vision techniques for the automatic identification of the target object, estimating its material and size, allowing the precise application of the required force to the target object. This novel approach enables users to efficiently grasp unknown objects with a significantly reduced failure rate. The vision-based method exhibits a grasp success rate of 87.5%, surpassing the baseline slip-grasp method’s 71.9%. These results underscore the effectiveness of our vision-based HMI in enhancing the grasp functionality of the exoskeleton glove.

Smart warehousing has been widely used due to its efficient storage and applications. However, the efficiency of transporting high-demand goods is still limited, because the existing methods lack path optimization strategies applicable to multiple scenarios and are unable to adapt conflict strategies to different warehouses. For solving these problems, this paper considers a multi-robot path planning method from three aspects: conflict-free scheduling, order picking and collision avoidance, which is adaptive to the picking needs of different warehouses by hierarchical agglomerative clustering algorithm, improved Reservation Table, and Dynamic Weighted Table. Firstly, the traditional A* algorithm is improved to better fit the actual warehouse operation mode. Secondly, the reservation table method is applied to solve the head-on collision problem of robots, and this paper improves the efficiency of the reservation table by changing the form of the reservation table. And the dynamic weighted table is added to solve the multi-robot problem about intersection conflict. Then, the HAC algorithm is applied to analyse the goods demand degree in current orders based on historical order data and rearrange the goods order in descending order, so that goods with a high-demand degree can be discharged from the warehouse in the first batch. Moreover, a complete outbound process is presented, which integrates HAC algorithm, improved reservation table and dynamic weighting table. Finally, the simulation is done to verify the validity of the proposed algorithm, which shows that the overall transit time of high-demand goods is reduced by 21.84% on average compared to the “A* + reservation table” algorithm, and the effectiveness of the solution is fully verified.

In recent years, the integration of large language models (LLMs) has revolutionized the field of robotics, enabling robots to communicate, understand, and reason with human-like proficiency. This paper explores the multifaceted impact of LLMs on robotics, addressing key challenges and opportunities for leveraging these models across various domains. By categorizing and analyzing LLM applications within core robotics elements—communication, perception, planning, and control—we aim to provide actionable insights for researchers seeking to integrate LLMs into their robotic systems. Our investigation focuses on LLMs developed post-GPT-3.5, primarily in text-based modalities while also considering multimodal approaches for perception and control. We offer comprehensive guidelines and examples for prompt engineering, facilitating beginners’ access to LLM-based robotics solutions. Through tutorial-level examples and structured prompt construction, we illustrate how LLM-guided enhancements can be seamlessly integrated into robotics applications. This survey serves as a roadmap for researchers navigating the evolving landscape of LLM-driven robotics, offering a comprehensive overview and practical guidance for harnessing the power of language models in robotics development.

Recognizing threats is a vital ability in social interactions across the animal kingdom. Yet, the role of specific perceptual elements, especially the head-body silhouette in aversive situations, remains relatively unexplored. In our study, we investigated the modulation of defensive behavior in rats facing a four-wheeled robot designed to simulate a natural predator. The robot featured an inflatable top allowing instant changes in appearance. In Experiment 1, rats encountered the head-inflatable robot (HEAD) in two sessions – a training session, where the rats were sequentially chased by the robot in both head-deflated (HEAD-Off) and head-inflated (HEAD-On) states, and a test session with a stationary HEAD-On or HEAD-Off robot 3 weeks later to assess long-term behavioral changes. The rats displayed reduced velocity and exploration in the center area during the HEAD-On phase of the training session. During the test session, the rats maintained a greater distance from the stationary HEAD-On robot than from the HEAD-Off robot, indicating sustained alertness based on the memory of the previous threat encounter. In Experiment 2, an identical procedure with the body-inflatable robot (BODY) was conducted. No significant differences emerged between BODY-On and BODY-Off conditions, except for a slight reduction in movement velocity during the BODY-On phase in the training session. Considering the substantial difference in behavioral reactions to HEAD-On versus BODY-On robots, we concluded that the emergence of head-like component in a chasing robot produced heightened vigilance and alertness. Since the two types of robots adopted a minimal design and differ only by the position of the inflatable top portion, our findings highlight the significant impact of a clearly recognizable head-like component in a threat encounter. The head-body silhouette provides a key perceptual framework for designing a social robot, with implications for both animal-robot and human–robot interactions.

In human communication, people often turn and gaze at a specific person in a crowd to signal their intention to interact with them. Similarly, it has been proposed that robots should also use social cues such as facing their human interaction partner during human–robot interaction tasks. This study introduces an initiatively interactive pose control (IPC) framework that allows a robot to face its task-relevant human interaction partner and proactively use this social cue while carrying out desired actions based on the ongoing task state. The IPC framework integrates a task planning module and a 3D identity recognition module. The task planning module can generate task states that include information about the desired human interaction partner’s name and the expected actions of robots, including social cues. The 3D identity recognition module implemented in the IPC framework can identify potential human interaction partners and estimate their pose in relation to the robot. The relative pose serves as the control parameter for orienting the robot toward the selected human interaction partner. The experimental results show that the IPC framework achieves a relative pose estimation error ranging from 0.04 to 6.56 degrees, which signifies a substantial enhancement compared to traditional sound source localization methods. Moreover, experiments also demonstrate that the robot can proactively turn toward the interaction partner and execute expected actions using the IPC framework. In conclusion, this paper introduces a new protocol for interactive pose control, enabling robots to actively select their human interaction partners and exhibit social cues and associated interaction actions.

Flexible dielectric elastomeric actuators (DEAs) have become significant in soft robots with intelligent systems. They overcome the shortcomings of traditional rigid systems, thereby expanding their applications in wearable devices. However, existing soft robot end-effectors have limited grasping adaptability and often require a complex coupling of sensors and control algorithms to achieve application data-driven smart grasping. This complexity significantly increases manufacturing costs and design difficulties. In this context, we present a simple, adaptive, and versatile double-finger soft gripper (DFSG) driven by a conical DEA to achieve compliant grips. The DFSG consists of three main parts: a conical actuator, clamp, and force transmission mechanism. Initially, we optimize the output performance of the conical actuator by tailoring its geometric structure, preload force, and bias voltage. The DFSG exploits the tapered actuator's characteristic of large vertical displacement (i.e., large input force) by utilizing the efficient displacement amplification function (up to 9 times) of the designed force transmission mechanism. It converts the input force in the vertical direction into a gripping force in the horizontal direction. As a result, the developed DFSG can easily grasp not only regular and stiff objects but also challenging objects such as small, irregular, soft, or squeezable items. Notably, it can clamp up to 14.5 times its own weight with just one layer of DEA. This work provides guidance for designing soft grippers with adaptive and high reliability, offering a promising avenue for the advancement of soft robotic systems.

Abstract

Autonomous battery charging is crucial for mobile service robots in human-center indoor environments, enabling them to extend operational hours and coverage without human assistance. This paper presents an innovative approach for mobile service robots to charge their batteries using standard wall outlets, introducing no additional maintenance cost and requiring no modification to environments. A portable self-charging device, equipped with cameras, a force sensor, and a 2-degree-of-freedom end-effector carrying a standard 3-pin 120V power plug, is attached to an existing mobile robot. The robot identifies a wall outlet and navigates to it using an onboard depth camera. It inserts the plug into the wall outlet while the vision is obstructed. The plug-insertion operation is guided by a control policy that was trained by a simulation model using a deep reinforcement learning technique. This approach achieved a success rate of nearly (90%) in experiments of inserting a power plug into a wall outlet. It eliminates the need of an installed docking station for autonomous charging or human plugging-in for manual charging.

Graphical Abstract

Compared to wheeled and tracked robots, hexapod robots have higher adaptability and higher flexibility in complex terrains. With various gaits, hexapod robots can fulfill different needs better. Existing researches mainly focused on three common gaits, they are single-leg swing gait, wave gait, and tripod gait. Instead of directly planning gaits with swarm intelligence algorithms (SIA), a gait planning method for hexapod robots named finite incremental state machine (FISM) is proposed. FISM focuses on four incremental states between two adjacent gaits of the robot, which greatly reduces the complexity of the gait planning algorithm so that gait planning with SIA is simplified to set the optimal transfer conditions of FISM. In addition, after comparing five optimization algorithms, the whale optimization algorithm (WOA) can set the optimal transfer conditions of FISM. The computer simulation shows WOA-FISM can plan various gaits, finally, a real robot test verifies the effectiveness of various gaits.

Lane-change maneuvers are a critical aspect of autonomous vehicles operation, but executing them efficiently and safely in the presence of other vehicles with varying driving behaviors, influenced by drivers’ emotions, poses a significant challenge. This paper presents a novel decision-making framework with trajectory generation and control algorithm, which considers the emotion-induced driving behavior of other vehicles’ drivers to perform safe and efficient lane-change maneuvers. The algorithm generates smooth trajectory candidates based on the position and velocity of other vehicles, selecting the most efficient and safest option. The control system tracks the generated lane-change trajectory, allowing the autonomous vehicle to pass the other vehicle if the driver is in a “happy,” “calm,” or “neutral” emotional state, exhibiting cautious behavior such as maintaining or reducing speed. Conversely, if the other vehicle’s driver is in an “angry” or “unpleasant” emotional state, causing aggressive behavior like accelerating and not allowing the autonomous vehicle to pass, the control system ensures the autonomous vehicle stays on its previous lane. Simulation and experimental results demonstrate that the proposed algorithm enables autonomous vehicles to perform lane-change maneuvers safely and efficiently in the presence of the other vehicle’s driver’s emotions, mitigating collisions. This proposed algorithm represents a significant step toward enabling autonomous vehicles to navigate complex traffic scenarios involving other vehicles with varying driving emotions.