In a groundbreaking development, researchers from Seron Electronics and the Massachusetts Institute of Technology (MIT) have unveiled an important advancement in the field of robotics, as detailed in a recent article in Advanced Materials Technologies. This collaboration has led to the enhancement of control systems for thermal artificial muscles, known for their outstanding power and energy density.
Central to this innovation is the integration of a robust machine-learning controller, that permits precise position control of these muscles without relying on external sensors. The machine learning model, particularly an ensemble encoder-style neural network, excels in translating desired displacement trajectories into the necessary power inputs. This development effectively equips the artificial muscles with a form of 'muscle memory,' a feature that is set to revolutionize their application in robotics and beyond.
A crucial component in this advancement is the Seron Electronics' SE Power Platform. This programmable power platform has been instrumental in providing constant power to the actuator and in seamlessly exciting the actuator with the power profile generated by the machine learning model. The integration of the SE Power Platform has been pivotal in achieving the remarkable precision and efficiency of these enhanced thermal artificial muscles.
This leap in technology heralds a new era in robotics, opening doors to enhanced functionality and broader integration of artificial muscles in various applications. The collaborative efforts of Seron Electronics and MIT researchers have paved the way for more sophisticated, efficient, and adaptable robotic systems.
Link to the research work: https://doi.org/10.1002/admt.202301769
Link to the SE power platform: Seron Electronics: Programmable Power Platform