Robotics and Autonomous Systems最新文献

This paper uses robot programming techniques, such as Deep Q Network, Artificial Neural Network, and Artificial Deep Q Network, to address challenges related to controlling robotic arms through demonstration learning. Static and dynamic states of the subjects were the subjects of experiments. Each method's classification accuracy process success values and experimental condition combination were evaluated. The DQN method demonstrated favourable classification accuracy outcomes, achieving an Accuracy value of 0.64 for the fixed dice and 0.52 for the moving dice. The Response value was 0.51 for the fixed dice and 0.41 for the moving dice, indicating a moderate level. The ANN method demonstrated lower accuracy, with Accuracy values of 0.59 and 0.56 and Response values of 0.61 and 0.58, respectively. The ADQN method demonstrated superior outcomes, with Accuracy values of 0.66 and 0.59 and Response values of 0.67 and 0.61. During the initial learning iterations, ADQN demonstrated the highest success rate at 33.67 %, whereas DQN and ANN achieved 28.39 % and 20.13 % success rates, respectively. As the number of iterations increased, all methods demonstrated improvement in their results. ADQN maintained a high success rate of 97.59 %, while DQN and ANN attained 82.16 % and 88.66 %, respectively. As the number of iterations increases, the results of all methods improve, but the success rate of the Artificial Deep Q Network remains high. As the number of iterations increases, both Deep Q Network and Artificial Neural Network demonstrate the potential to achieve good results. Overall, the findings support the efficacy of robot programming techniques that incorporate demonstration learning. The Artificial Deep Q Network is the most successful and fast-converging method suitable for various robot control tasks. These findings provide a foundation for future research and large-scale, comprehensive learning applications for complex rot control.

Cooperative, connected and automated mobility (CCAM) is one of the next big steps in the automotive industry. Thanks to recent improvements in Advanced Driver Assistance Systems, and novel methods for automating vehicles, more safe and efficient transport mechanisms have been achieved. Current vehicles are already connected devices, and communications between vehicles, infrastructure and other road users will allow traffic agents to share information and use it to coordinate their actions. The full integration between cooperation, connectivity, and automation technologies entail an important achievement to improve road safety, traffic efficiency and comfort of driving. To approach this goal, the main contributions of this work propose a new distributed mission system based on Advanced Behavioral Points (ABP). That is, based on relevant points inside a plan which store a collection of predefined tasks that operate at the high level layer of an automated and connected vehicle to coordinate behaviors when connecting to critical emplacements like junctions and roundabouts. This approach, which has been tested in the simulation environment of Carla, provide a collaboration stack between the traffic infrastructure and the ego vehicle so as to cope with actual problems such as traffic congestion and road accidents.

This paper proposes a visual predictive control solution adapted to mobile manipulators and able to cope with several issues related to visibility, manipulability, and stability. To address these problems, the proposed strategy relies on (i) the use of two complementary cameras, (ii) the definition of a cost function depending on both the vision-based task and the manipulability, (iii) the integration of time-varying constraints allowing to prioritize the former against the latter. The strategy has been analyzed through simulation using ROS and Gazebo and implemented on our TIAGo robot. The obtained results fully validate the proposed approach.

Many daily tasks exhibit a periodic nature, necessitating that robots possess the ability to execute them either alone or in collaboration with humans. A widely used approach to encode and learn such periodic patterns from human demonstrations is through periodic Dynamic Movement Primitives (DMPs). Periodic DMPs encode cyclic data independently across multiple dimensions of multi-degree of freedom systems. This method is effective for simple data, like Cartesian or joint position trajectories. However, it cannot account for various geometric constraints imposed by more complex data, such as orientation and stiffness. To bridge this gap, we propose a novel periodic DMP formulation that enables the encoding of periodic orientation trajectories and varying stiffness matrices while considering their geometric constraints. Our geometry-aware approach exploits the properties of the Riemannian manifold and Lie group to directly encode such periodic data while respecting its inherent geometric constraints. We initially employed simulation to validate the technical aspects of the proposed method thoroughly. Subsequently, we conducted experiments with two different real-world robots performing daily tasks involving periodic changes in orientation and/or stiffness, i.e., operating a drilling machine using a rotary handle and facilitating collaborative human–robot sawing.

This article discusses a novel aerial robot architecture that overcomes the underactuation of conventional multirotor systems without adding dedicated rotor tilting actuators. The proposed system is based on four quadrotors cooperatively carrying a central body to which they are attached through passive universal joints. While conventional parallel axis multirotors are underactuated, the proposed mechanism makes the system overactuated, enabling independent position and orientation control of the main body. This implies that the payload can be carried in the minimum drag orientation, it enables take-off and landing on inclined surfaces and it provides thrust-vectoring capabilities to the system, leading to high control authority. A detailed dynamic model is derived making use of Lagrangian formalism and a hierarchical control law based on such model is proposed to stabilize the system. This control law is designed to ensure good tracking while minimizing power consumption. The proposed control law and the capabilities of the architecture are evaluated in simulation and in outdoor experimental flights, where the aerial robot shows autonomous tracking of the six degrees of freedom (DoF) of the main body, an inherently unfeasible maneuver for conventional underactuated multirotors.

In the real world, multi-robot systems need to deal with on-the-fly (runtime) arrivals of new sets of tasks. This entails repeated adjustments of their current task allocations to include the newer ones while also ensuring that the overall performance does not degrade. This paper proposes a decentralized and distributed dynamic task allocation algorithm to handle this issue in a multi-robot scenario. The proposed work provides a conflict-free allocation of a set of tasks constituting a job to robots and minimizes the total execution time. These jobs can comprise multiple independent and/or dependent tasks or a combination thereof, which are injected on-the-fly into a network of robots. The dependent tasks of a job are related by precedence constraints that specify the ordering or dependencies between pairs of tasks. The work also describes a decentralized adaptive energy threshold mechanism for determining whether or not a robot needs to visit a battery stockpile after the execution of a task. Conflicting task selections among the robots in this decentralized set-up are resolved using mobile agents during runtime. Apart from allocating tasks to the robots, these mobile agents exploit the benefits of centralized and decentralized systems and provide an advantage over auction-based task allocation algorithms. The proposed algorithm takes into consideration the energy requirements, both during the task allocation process and actual execution. The proposed algorithm also caters to strategies to deal with delays caused by obstacles and congestion during the actual execution of the tasks. Experiments conducted using Webots, an open-source robot simulator, and Tartarus, a multi-agent platform, authenticate the efficacy of the proposed algorithm compared to other prominent task allocation algorithms in terms of minimization of average waiting time, total task allocation time, total job allocation time, and total execution time of an experiment.

The integration of robotics in nursing is a significant shift in healthcare, driven by the aging global population and the increasing demand for care. Robots in nursing can handle less technical tasks such as patient transport and rehabilitation activities. This support allows caregivers to focus on less strenuous nursing duties and more direct patient care. Human-Robot Interaction (HRI) plays an important role in this challenging context. In this research, we present an autonomous hospital transport system based on the ROS 2 framework, focusing on enhancing HRI in the healthcare environment. It encompasses the development of a control architecture for autonomous robot behavior, the implementation of machine learning for emergency detection, and the creation of a user-friendly interface for both patients and staff. The proposed concepts were validated in real-world scenarios in three different hospitals in Germany. This not only demonstrates the practical application of this system but also shares insights and methods, encouraging further advancement in the field of healthcare robotics.

Stability of obstacle–crossing and structural optimization are important issues in the research of tracked mobile robots. In this paper, in order to fully understand the obstacle–surmounting ability of the robot, the relationship between the position of the center of gravity and the posture of the front and rear swing arms is analyzed. Based on the motion mechanism of the robot crossing obstacles, the geometric model and the dynamic model are established for the key states in the obstacle crossing process. Based on these models, a multi-objective optimization problem for the maximum obstacle–crossing height and minimum driving torque is established during the obstacle crossing process of the robot, which must meet geometric, slip, and stability constraints. To effectively handle the optimization problem of tracked mobile robots, an improved non–dominated sorting genetic algorithm with elite strategy version II based on adaptive genetic strategy (NSGA-II-AGS) is proposed in this paper. Some meaningful relationships between the objective function and the design variables are obtained through sensitivity analysis. Finally, the robot's obstacle-crossing ability was verified through virtual simulation and prototype experiments. These excellent performances enable the proposed NSGA-II-AGS to be qualified for dealing with the multi-objective optimization problem.

The ability of robots to navigate through doors is crucial for their effective operation in indoor environments. Consequently, extensive research has been conducted to develop robots capable of opening specific doors. However, the diverse combinations of door handles and opening directions necessitate a more versatile door opening system for robots to successfully operate in real-world environments. In this paper, we propose a mobile manipulator system that can autonomously open various doors without prior knowledge. By using convolutional neural networks, point cloud extraction techniques, and external force measurements during exploratory motion, we obtained information regarding handle types, poses, and door characteristics. Through two different approaches, adaptive position-force control and deep reinforcement learning, we successfully opened doors without precise trajectory or excessive external force. The adaptive position-force control method involves moving the end-effector in the direction of the door opening while responding compliantly to external forces, ensuring safety and manipulator workspace. Meanwhile, the deep reinforcement learning policy minimizes applied forces and eliminates unnecessary movements, enabling stable operation across doors with different poses and widths. The RL-based approach outperforms the adaptive position-force control method in terms of compensating for external forces, ensuring smooth motion, and achieving efficient speed. It reduces the maximum force required by 3.27 times and improves motion smoothness by 1.82 times. However, the non-learning-based adaptive position-force control method demonstrates more versatility in opening a wider range of doors, encompassing revolute doors with four distinct opening directions and varying widths.

This paper presents a novel dynamic motion planner designed to provide safe motions in the context of the Smart Autonomous Robot Assistant Surgeon (SARAS) surgical platform. SARAS is a multi-robot autonomous platform designed to execute auxiliary tasks in Minimally Invasive Surgeries (MIS) with a high degree of autonomy. The development of robotic systems with a high level of autonomy and reliability requires to perceive the workspace and human actions, to contextualize them with the surgical workflow, and, finally, plan and dynamically control the required motions. The autonomous control relies on a multi-level hierarchical Finite State Machine (hFSM) that decides and supervises all robot actions and their transitions. This approach requires multi-granularity decomposition of the surgical procedure and defines different motion profiles to preserve and safely interacts with the patients’ anatomy. The motion planner is developed under the minimally invasive surgery context since it is an extreme use case where the environment is complex, dynamic and unstructured. Moreover, in the SARAS platform the autonomous robots share workspace as well as collaborate with other human-guided robotic instruments. This creates an even more complex working environment and defines a set of hierarchical relationships in which auxiliary instruments have a lower priority. The presented motion planner acts at two levels: Global and Local. The Global Planner generates an initial spline-based trajectory that, defined by a set of Control Points, follows a certain profile determined by the ongoing surgical action and the interaction with the patient’s anatomy. Then, during the execution of the motion, the Local Planner observes the workspace (anatomy and other tools) and applies different virtual potential fields to the control points to dynamically modify their position to avoid potential collisions or tool blocking while maintaining trajectory coherence. At this level, it reactively modifies the trajectory between the tool position and the next control point applying Dynamical Systems based obstacle avoidance. This approach ensures collision free connections between the spline control points. The proposed motion planner is validated in a realistic surgical scenario. The experimental results are analysed from data collected during various Robotic-Assisted Radical Prostatectomy surgeries on manikins, performed with the SARAS SOLO-SURGERY platform: the main surgeon teleoperates a daVinci Research Kit and two robotic arms autonomously perform different auxiliary surgical tasks.