A deep learning and model predictive control framework to control quadrotors and agile robots

As technology populations and advances, there has been a significant push to create robots and robotic agents that are capable of acting as independent entities within controlled and unstructured environments. One of the key areas of advances for robotic agents has been in the development of algorithms for controlling the movements of these robotic agents.

Model predictive control (MPC) algorithms are a type of algorithm designed to control the movement of robotic agents. These algorithms are used to optimise and control the behaviour of a robotic agent in the short term towards a given goal, while also satisfying any limitations or constraints that may be put in place.

MPC algorithms are widely used in robotics, especially in robotics that must interact with humans. For example, if a robotic arm is required to move something in a room but must also ensure it does not crash into furniture and/or other obstacles along the way then an MPC algorithm could be used to ensure the arm is guided to its destination safely.

MPC algorithms have been developing rapidly over the past few years and have been used to create robotic agents that can interact effectively and safely with humans and the environment around them. This has been particularly useful in industrial settings where robotic agents are required to interact with moving objects and resources, such as in warehouses, factories, and construction sites.

The advances in MPC-controlled robots could revolutionise the way robotic agents interact with each other and humans. As these algorithms get more advanced, robot-human collaborations are likely to become more common, with robots acting as effective assistants that can help us achieve our goals more effectively and more safely.






Leave a Reply

Your email address will not be published. Required fields are marked *