Advancements and Applications of Neural Networks: From Automation to Human-Machine Interaction

The field of neural networks has experienced significant advancements in recent years, offering various advantages in modeling. These advantages, such as improved accuracy and efficiency, have resulted in successful applications across multiple domains, including robotics, smart manufacturing, Internet of Things (IoT), healthcare automation and many others. Furthermore, recent research has demonstrated the potential of neural networks in generating control commands, either by combining traditional control algorithms like fuzzy logic, adaptive control, Bayes decision control, and Gaussian decision control, or by employing end-to-end schemes. These advantages have been further validated in decision-making processes, highlighting the capacity of neural networks to operate autonomously.

However, these systems can encounter challenges due to limited environmental awareness and uncertainties. To address these challenges, the integration of human experience into autonomous systems has emerged as a promising paradigm. The interaction between humans and machines, whether through physical collaboration or teleoperation, presents both new possibilities and challenges that require further exploration. In the case of teleoperation, there are novel and specific challenges (such as time delays between human commands and robot actions, communication quality, human intention estimation, provision of hybrid sensory information feedback, and human perception augmentation and human performance evaluation) to address that can impact user training and adoption of these systems.

The special session will cover a range of topics (keywords), including but not limited to:

• Robot learning for various tasks including perception, object recognition, motion planning, and control.

• Intelligent manufacturing including equipment failures prediction, anomalies detection, production schedules optimization, quality control, predictive maintenance, and energy consumption optimization.

• Neural network-based methodologies in additive manufacturing, 3D printing, processing, and joining.

• Communication strategies in autonomous systems and teleoperation.

• Integrating intelligent systems with cyber-physical systems for enhanced performance and reliability.

• Machine/deep learning-based human intention recognition/estimation in human-machine/robot collaboration.

• Multilateral teleoperation control, coordination, and remote arbitration techniques with reinforcement learning algorithms.

• Machine learning for sensory fusion and perception enhancement during remote communication.

• Neural network-based user training and interface design for intuitive teleoperation

Submission from: https://edas.info/newPaper.php?c=31628&track=122635