this paper deals with collision avoidance for an autonomous vehicle (AV) using a model-free Reinforcement Learning (RL) algorithm rooted in the actor-critic paradigm. To achieve this objective, the actor network (AN) ...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
this paper deals with collision avoidance for an autonomous vehicle (AV) using a model-free Reinforcement Learning (RL) algorithm rooted in the actor-critic paradigm. To achieve this objective, the actor network (AN) has to generate a collision-free path for an autonomous robot from a start to an end position as well as to follow this desired path accurately. Within this framework, the actor provides a sequence of input signals for the underlying velocity controllers of the robot drives. To accomplish this purpose for a large number of obstacles, it turns out to be essential to sort the algorithm's input vector regarding the smallest Euclidean distance between an obstacle and the agent as well as to consider the robot's relative direction. In a first step, the training of the agent is performed in a simulated environment. the second step involves the successful experimental validation of the trained AN on a TurtleBot 3 Burger (TB3B) - a test platform for autonomous robots.
this paper proposes a reinforcement learning (RL) framework for controlling and stabilizing the Twin Rotor Aerodynamic system (TRAS) at specific pitch and azimuth angles and tracking a given trajectory. the complex dy...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
this paper proposes a reinforcement learning (RL) framework for controlling and stabilizing the Twin Rotor Aerodynamic system (TRAS) at specific pitch and azimuth angles and tracking a given trajectory. the complex dynamics and non-linear characteristics of the TRAS make it challenging to control using traditional control algorithms. However, recent developments in RL have attracted interest due to their potential applications in the control of multirotors. the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm was used in this paper to train the RL agent. this algorithm is used for environments with continuous state and action spaces, similar to the TRAS, as it does not require a model of the system. the simulation results illustrated the effectiveness of the RL control method. Next, external disturbances in the form of wind disturbances were used to test the controller's effectiveness compared to conventional PID controllers. Lastly, experiments on a laboratory setup were carried out to confirm the controller's effectiveness in real-world applications.
this paper proposes a two-layer robust control framework for three-axis attitude reference tracking while maintaining requirements on the spacecraft's actuators and motion profile in the presence of uncertainty. T...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
this paper proposes a two-layer robust control framework for three-axis attitude reference tracking while maintaining requirements on the spacecraft's actuators and motion profile in the presence of uncertainty. the approach builds upon a nonlinear Lyapunov controller in the inner-loop that seeks to stabilize the nonlinear spacecraft equations of motion. the resulting closed-loop system is augmented by an outer-loop Tube-Based Model Predictive controller (TBMPC) that leverages the benefits of robust constraint satisfaction. To verify the performance of the control framework, a Monte-Carlo simulation with 150 samples subject to varying uncertainty is conducted in MATLAB. Key findings include that TBMPC outperforms a conventional Model Predictive controller (MPC) and maintains the actuator and performance requirements for any admissible disturbance realization. Overall, the results presented in this work prove online applicability of TBMPC for the specified high-dimensional pointing scenario.
this work is set to design pose-free visual feedback control laws. Our contribution is linked to improving 3D posefree visual servoing strategy using 2D predictions of visual features' trajectories. Given a set ri...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
this work is set to design pose-free visual feedback control laws. Our contribution is linked to improving 3D posefree visual servoing strategy using 2D predictions of visual features' trajectories. Given a set rigid body characterized by a set of 3D point features, our proposed visual servoing strategy enhanced with predictive elements is based on computingthe following sample time positions of the point features given the known camera velocity from the previous time moment. In this way, the architecture decides to discard the point feature before exiting the field of view. thus, any unnecessary computations are removed, and the camera's motion is ensured to be more natural. the simulation results emphasize the potential of the new visual control architecture.
the generator blocks with which the thermoelectric plants are equipped are structures that must satisfy specific operating regimes imposed by the working conditions of the national energy system. To supervise and adju...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
the generator blocks with which the thermoelectric plants are equipped are structures that must satisfy specific operating regimes imposed by the working conditions of the national energy system. To supervise and adjust these structures running mode, continuous monitoring, and specific recording of the parameters afferent to their functioning regimes are required. Since the plant organization and the energy system performance may differ, customized implementations for specific electric power architecture may provide the required monitored information in the optimal way. the present work details the development process and the means of use for a customized data acquisition architecture which meets the monitoring and recording requirements specific to generator block operation. this application currently runs in thermoelectric power plants in Romania. the hardware structure is managed through a software suite, implemented as a handy tool for the monitoring and recording process, through features such as record storing and report generation.
Object-grasping is one of the most common tasks in daily life. However, this task is very challenging for elderly human subjects with weak or dysfunctional hand. therefore, hand assistive device is necessary to enhanc...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
Object-grasping is one of the most common tasks in daily life. However, this task is very challenging for elderly human subjects with weak or dysfunctional hand. therefore, hand assistive device is necessary to enhance the grasping performance. For the improvement in grasping assistance task, force control strategy along with exoskeleton's design, is one of the major objectives in assistive technology. this study proposes a robust sliding mode force controller scheme for different-sized object-grasping tasks. To verify the efficacy of the proposed controller during object-grasping task, constant grasping experiment is performed on two elderly human subjects. Furthermore, we have incorporated real-time external disturbance rejection experiment to establish the robustness of the proposed force control scheme. the extensive grasping experiments involving human subjects show that the proposed robust force control strategy withthe designed two-fingered exoskeleton can be preferable for performing object-gasping task in hand assistive technology.
Autonomous Mobile Robots (AMRs) are essential in various industries due to their efficacy and their usefulness in securing the transportation of goods. these robots autonomously execute tasks in intra-logistics operat...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
Autonomous Mobile Robots (AMRs) are essential in various industries due to their efficacy and their usefulness in securing the transportation of goods. these robots autonomously execute tasks in intra-logistics operations, spanning manufacturing, warehousing, and healthcare settings. Moreover, AMRs can interact with humans as collaborative assistants. this paper proposes a longitudinal coalitional control (CC) approach for an interconnected AMRs group. the CC method ensures that the AMRs are controlled so that, both position and velocity errors are minimised, thereby maintaining the shape of the group. Additionally, the method allows for dynamic communication topology switching by enabling or disabling communication links to reduce communication usage. the results show that the coalitional algorithm with switching topologies yields a better cost performance compared to individual topologies, proving the efficacy of the CC method in achieving satisfactory performances with minimal communication usage.
Emulsification processes show a plethora of use cases in different industries. the complexity and intransparency of many emulsion systems make it hard to apply classic control approaches. the operation of these system...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
Emulsification processes show a plethora of use cases in different industries. the complexity and intransparency of many emulsion systems make it hard to apply classic control approaches. the operation of these systems therefore often diverts towards a manual open-loop control. While population balance models (PBM) have been explored for multiple decades, they are rarely used in practice for closed-loop control due to the high computational effort. For this purpose a datadriven modeling approach specifically tailored to the control of complex emulsification devices is introduced. A new simplified description scheme of particle size distributions in combination with Gaussian process regression on a reasonably sized dataset can predict the system change. It additionally gives a useful measure of uncertainty for the predicted change, which is propagated onto the discrete distribution description. the concept is proven with leave-one-out cross-validation, before showing its potential in a model predictive control (MPC) simulation.
Efficient operation and maintenance of wastewater treatment plants (WWTPs) are essential for safeguarding public health and the environment. the emergence of mechanical faults within complex systems can lead to disrup...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
Efficient operation and maintenance of wastewater treatment plants (WWTPs) are essential for safeguarding public health and the environment. the emergence of mechanical faults within complex systems can lead to disruptions, increased operational costs, and environmental risks. As the world moves towards a digitally connected and sustainable future, the development of Deep Learning (DL) tools for fault detection and isolation (FDI) in wastewater treatment processes is expected to become paramount. therefore, in this study, we developed two neural models, a Feedforward Neural Network (FFNN) and a Long Short-Term Memory (LSTM), to address the detection of mechanical faults such as bias, stuck, spikes, and precision degradation of the Dissolved Oxygen (DO) sensor. the classification results showed remarkable accuracy performances during testing: for Dataset 1, FFNN achieved 96.56%, while LSTM reached 99.36%;and for Dataset 2, FFNN achieved 99.36%, and LSTM reached 99.57%.
this paper analyses models that can realistically map the grayscale image domain to the color image domain. We thus focus our attention on a CycleGAN (Cycle-consistent Generative Adversarial Network) neural network ar...
详细信息
ISBN:
(纸本)9798350364309;9798350364293
this paper analyses models that can realistically map the grayscale image domain to the color image domain. We thus focus our attention on a CycleGAN (Cycle-consistent Generative Adversarial Network) neural network architecture, which proves to have good results in the area of translation between domains, and implicitly in the application of image coloring. In addition to exploring CycleGAN for image colorization, this study introduces novel techniques aimed at enhancing its performance. We incorporate a color distribution loss term to ensure reliable color mapping, effectively addressing discrepancies in color distribution between domains. Moreover, an in-depth analysis of generator errors is conducted, unveiling critical insights into model limitations. Leveraging this analysis, we propose a correction network to generate residual images, facilitating more accurate colorization. through rigorous experimentation across diverse datasets, our approach showcases a remarkable capacity to produce lifelike and coherent colorizations.
暂无评论