Journal international des progrès technologiques
Libre accès
ISSN: 0976-4860
ISSN: 0976-4860
Samantha Williams
Major difficulties and challenges of modern robotics systems are around how to give robots the ability to learn and make decisions for themselves. The visual servoing control approach is a common robotic system strategy for perceiving the environment through vision. New robotic systems can be guided by the vision to execute more sophisticated tasks in more complex working contexts. This survey attempts to describe current learning-based algorithms, particularly those that combine with Model Predictive Control (MPC) in visual servoing systems, as well as provide some pioneering and advanced references with numerical simulations. The impact of classical control tactics on robotic visual servoing systems is discussed, as well as generic modeling methods for visual servoing. The benefits of incorporating neural network- and reinforcement-learning-based algorithms into systems are explored. Finally, the future directions of robotic visual servoing systems are summarised and forecasted based on existing research progress and references.