University of California, Los Angeles (UCLA) mechanical engineers have developed a novel material that uses AI to learn behaviours over an extended period-of-time.
Benefits of the material in different industries
The material is made up of a structural system made-up of adjustable beams that can alter shape & behaviour over time. According to the study, this change is a response to dynamic situations. The research team revealed that this AI may be used to build buildings, planes, & imaging equipment.
“This research introduces and demonstrates an AI material that can learn to exhibit the desired behaviours & properties with increased exposure to ambient conditions,” said Jonathan Hopkins, the study’s lead researcher and mechanical and aerospace engineering professor at UCLA’s Samueli School of Engineering.
He stated that the principles utilised in this study are the same as those used in machine learning, giving the material the potential to adapt.
The study’s example indicates the potential for employing the material in aeroplane wings. The AI material may learn and adapt to take on the form of the wings. In order to increase flexibility and efficiency during a flight, this would take place based on the wind patterns.
The research team also discussed the advantages of incorporating the material into buildings to increase stability during earthquakes, hurricanes, and other natural disasters.
Creating the material
The study team developed the material using ideas from already-existing artificial neural networks (ANNs). The algorithms that drive machine learning are ANNs. The material mechanical neural network is referred to as (MNN). The MNN is made up of individually adjustable beams arranged in a triangle-shaped lattice structure. Each separate beam contains a “Voice coils, strain gauges, & flexures that enable the beam to modify its length, adapt to its changing environment in real time, and interact with other beams in the system”. This enables the material to keep its environmental flexibility.
Purpose of the voice coil, strain gauges and flexures
The paper discusses the function of flexures, strain gauges, and voice coils. The study claims that the voice coil, which derives its name from the fact that it was orignal used in speakers to convert magnetic fields into mechanical motion, “initiates fine-tuned compression or expansion in reaction to fresh forces applied to the beam,” according to the study.
“Data from the motion of the beam are collected by the strain gauge and used by the algorithm to control the learning behaviour. The flexures essentially act as flexible joints between the moving beams to connect the system.” All 3 work together to allow for elasticity & flexibility.
The completion includes an optimization algorithm that regulates the entire system, taking data from the strain gauges and creating stiffness values to control how the network should adapt. Controls the force to be applied. In order to verify the accuracy of the strain gauge system, cameras are also mounted on the system’s outer nodes.
The MNN system is currently the size of a microwave oven. The research team, on the other hand, wishes to simplify the concept so that thousands of the systems can be constructed on a much smaller scale and used to diverse tasks.
The team hopes to make the materials available for use in future shock-absorbing armour in addition to using them in aeroplanes & buildings.
The study was published today in the journal Science Robotics.