Robotics and Autonomous Systems最新文献

The flexible four-finger gripper, as a specialized robotic end-effector, is highly valued for its ability to passively adapt to the shape of objects and perform non-destructive grasping. However, the development of grasping detection algorithms for flexible four-finger grippers remains relatively unexplored. This paper addresses the unique characteristics of the flexible four-finger gripper by proposing a grasping detection method based on deep learning. Firstly, the Acute Angle Representation model (AAR-model), which is based on the structure of the flexible four-finger gripper and consists of grasp points and angles, is designed as the grasping representation model that reduces unnecessary rotations of the gripper and improves its versatility in grasping objects. Then, the Flexible Gripper Adaptive Attribute model (FGAA-model) is proposed to represent the grasping attributes of objects, calculate the grasp angles that meet the criteria of the AAR-model, and aggregate the AAR-models on the image data into a unified set, thereby circumventing the time-consuming process of pixel-level annotation. Finally, the Adaptive Grasping Neural Net (AGNN), which is based on Adaptive Feature Fusion and the Grasp Aware Network (AFFGA), is introduced by eliminating redundant network detection headers, fusing color and depth images as inputs, and incorporating a Series Atrous Spatial Pyramid (SASP) structure to produce more accurate grasp poses. Our method not only attains a remarkable accuracy of 97.62% on the Cornell dataset but also swiftly completes grasping detection within 25 ms. In practical robotic arm grasping tests, where a robot is outfitted with a flexible four-finger gripper, it successfully grasps unknown objects with a 96% success rate. These results underscore the reliability and real-time performance of our method, significantly enhancing the gripper's adaptability and precision when handling objects of varying sizes and shapes. This advancement provides a powerful technical solution for robots utilizing flexible four-finger grippers, enabling autonomous, real-time, and highly accurate grasping maneuvers. Moreover, it addresses the persistent challenge of the scarcity of efficient grasping detection techniques tailored for flexible four-finger grippers.

In this paper, a motion planning algorithm for floating planar under-actuated hyper-redundant snake robots is proposed. The presented algorithm generates locally optimal shape trajectories, i.e., continuous trajectories in the base space of the robot. Such shape trajectories produce a desired rotation of the snake robot, i.e., change in the uncontrolled orientation fiber variable. The proposed method formulates the motion planning problem as an optimization problem where the objective function could be defined to minimize various metrics, such as energy-based cost functions. Additionally, the proposed motion planning algorithm uses a heuristic to generate shape trajectories that avoid self-intersections and obstacle collision. Hence, the motion planning method generates shape trajectories that locally minimize user-defined cost functions and eliminate self-intersections or obstacle collision. The proposed gait generation method is validated using numerical simulations of five-link and seven-link snake robots.

Behavior trees (BTs) are increasingly popular in the robotics community. Yet in the growing body of published work on this topic, there is a lack of consensus on what to measure and how to quantify BTs when reporting results. This is not only due to the lack of standardized measures, but due to the sometimes ambiguous use of definitions to describe BT properties. This work provides a comprehensive overview of BT properties the community is interested in, how they relate to each other, the metrics currently used to measure BTs, and whether the metrics appropriately quantify those properties of interest. Finally, we provide the practitioner with a set of metrics to measure, as well as insights into the properties that can be derived from those metrics.

By providing this holistic view of properties and their corresponding evaluation metrics, we hope to improve clarity when using BTs in robotics. This more systematic approach will make reported results more consistent and comparable when evaluating BTs.

We live in a constantly changing world. For robots to fully operate in our world, they need to work in dynamic environments where objects are not fixed in place or may be moved by humans or other agents. This work is based on tactile sensing, as it enables sufficiently responsive robotic systems for contact-based tasks in dynamic environments. Our proposed approach is divided into two parts: (1) a way to perform object following using a shear controller that minimises tactile shear deformation and (2) a switching controller that alternates between the shear controller and a tactile exploration controller that enables contour-following of a moving object. We find that during the object-following task, the robot follows the moving object to sub-millimetre accuracy over a $\approx$72 mm range for 5 different velocities in 2D. The switching controller successfully performs 2D contour following on several moving objects at various object speeds whilst keeping an almost constant speed of exploration. We expect our method for minimising sensor deformation using a simple controller will generalise over different kinds of contact scenarios for moving objects. Moreover, the switching controller provides an architecture where velocity information of moving objects is fused with another controller thereby enabling a more holistic use of tactile information to empower robotic systems to perform complex tactile tasks.

The article presents a novel idea to construct a smooth navigation function for a wheeled robot based on grid-based search, that enables replanning in dynamic environments. Since the dynamic constraints of the robot are also considered, the navigation function is combined with the model predictive control (MPC) to guide the robot safely to the defined goal location. The main novelty of this work is the definition of this navigation function and its MPC application with guaranteed closed-loop convergence in finite time for a non-holonomic robot with speed and acceleration constraints. The navigation function consists of an interpolated potential function derived from the grid-based search and a term that guides the orientation of the robot on continuous gradients. The navigation function guarantees convergent trajectories to the desired goal, results in smooth motion between obstacles, has no local minima, and is computationally efficient. The proposed navigation is also suitable in dynamic environments, as confirmed by experiments with a Husky mobile robot.

The model-free controller is a powerful method because it does not require the knowledge of the robotic plant dynamic equation. Since most of the model-free controllers consider variants of the adaptive or reinforcement learning, one observer-based model-free controller could be a different alternative of interest in the researching community. In this study, an observer-based model-free controller is suggested for the perturbations estimation and attenuation in robotic plants, it contains the model-free observer and the model-free controller. The model-free observer based on the high gain method uses the bounds of the perturbations. The model-free controller based on the sliding mode method uses the bounds of the perturbations and gravity terms. The observer-based model-free controller is applied for the perturbations estimation and attenuation in a cylindrical robotic plant, and a scalar robotic plant.

The ability to avoid collisions with moving robots is critical in many applications. Moreover, if the robots have limited battery life, the goal is not only to avoid collisions but also to design efficient trajectories in terms of energy consumption and total mission time. This paper proposes a novel strategy for assigning turn angles for collision-free path planning in scenarios where a small team of robots cooperate in a certain mission. The algorithm allows each robot to reach a predetermined destination safely. It establishes consecutive, short time intervals, and at each interval, possible conflicts are solved centrally in an optimal manner. This is done by keeping constant speeds but generating a discrete set of possible directions for each robot, and solving efficiently the turn-angle allocation for a collision-free path that minimizes the path deviation from the shortest one. Due to the discretization, the final paths are not optimal, but the system can react to possible failures during execution, as conflicts are resolved at each time interval. Computational results and Software-In-The-Loop simulations are presented in order to evaluate the proposed algorithm. A comparison with a state-of-the-art approach shows that our algorithm is more energy-efficient and achieves lower mission completion time.

This article addresses the predefined-time trajectory tracking control of free-flying space manipulator (FFSM) subject to uncertainties and disturbances. First, a predefined-time nonsingular terminal sliding mode (NTSM) controller is developed with the high insensitivity to uncertainties and strong robustness against disturbances. The generalized attitude and angular velocity tracking errors under the predefined-time NTSM controller can stabilize to zero in predefined time. Then, an adaptive version of the predefined-time NTSM controller is presented. The parametric adaptation mechanism is incorporated to identify the square of the upper bound of the lumped unknown item. Thus, the predefined-time adaptive NTSM (ANTSM) controller is smooth with no obvious chattering phenomenon and can maintain the high tracking accuracy simultaneously. The generalized attitude and angular velocity tracking errors under the predefined-time ANTSM controller can stabilize to the minor bounded regions around zero in predefined time. Simulations are provided to demonstrate the exploited controllers.

Trunk compensation is a common behavior observed in stroke patients during rehabilitation, and it can hinder their recovery outcomes. To address this issue, we developed a new upper-limb rehabilitation robot that takes advantage of both end-effector and exoskeleton robots. Moreover, we propose a compensation-corrective adaptive control (CCAC) strategy, which employs an admittance model and incorporates two estimators. Specifically, the first estimator is designed to assess human intention, allowing for compliant human-robot interaction. The second estimator calculates dynamic assistance that adjusts for trunk compensation, utilizing two virtual forces applied to the hand and shoulder. Based on this novel CCAC strategy, the newly designed robot is capable of assisting upper limb movements and correcting compensatory postures simultaneously. Results indicate a significant reduction in trunk compensation across three types of reaching tasks when the robot provides assistance. Moreover, the CCAC strategy enhances upper-limb motor performance, resulting in reduced position errors and increased shoulder and elbow joint angles. These findings underscore the potential of the proposed CCAC strategy, combined with upper-limb exoskeleton robots, as a promising approach for correcting compensatory postures and optimizing the advantages of robotic stroke rehabilitation.

Ego-motion estimation plays a critical role in autonomous driving systems by providing accurate and timely information about the vehicle’s position and orientation. To achieve high levels of accuracy and robustness, it is essential to leverage a range of sensor modalities to account for highly dynamic and diverse scenes, and consequent sensor limitations.

In this work, we introduce TEFu-Net, a Deep-Learning-based late fusion architecture that combines multiple ego-motion estimates from diverse data modalities, including stereo RGB, LiDAR point clouds and GNSS/IMU measurements. Our approach is non-parametric and scalable, making it adaptable to different sensor set configurations. By leveraging a Long Short-Term Memory (LSTM), TEFu-Net produces reliable and robust spatiotemporal ego-motion estimates. This capability allows it to filter out erroneous input measurements, ensuring the accuracy of the car’s motion calculations over time. Extensive experiments show an average accuracy increase of 63% over TEFu-Net’s input estimators and on par results with the state-of-the-art in real-world driving scenarios. We also demonstrate that our solution can achieve accurate estimates under sensor or input failure. Therefore, TEFu-Net enhances the accuracy and robustness of ego-motion estimation in real-world driving scenarios, particularly in challenging conditions such as cluttered environments, tunnels, dense vegetation, and unstructured scenes. As a result of these enhancements, it bolsters the reliability of autonomous driving functions.