Prof. Carmel Majidi, Artificial Muscle and Artificial Skin
Posted on January 12th, 2015
01/12/2015 @ Denton’s 1221 6th Ave, NY
Simone Braunstein started the meeting with a brief review of the articles from the previously meeting and an invitation to participate in a competition announced on the SoftRoboticsToolkit web page. Following this introduction, Prof. Carmel Majidi @CarnegieMellon University talked about research his lab is conduction.
He contrasted soft vs hard robotics and noted the importance of research in soft robotics in terms of safety, comfort, and a good match with biological functions. He also emphasized the importance of the robot’s surface being : softness, stretchable, light weight, non-toxic/biocompatible.
Carmel then talked about the challenges of creating muscle and skin
- Artificial muscle. Methods to create muscle function include elastic tubing, embedded fibers and mechanical coupling with an inflate diaphragm. He talked about promising methods including
- Pneumatics g. Baldwin Actuator – usually not load bearing. Inflate to actuate. One of the challenges of this approach is that process is driven by bulky (and hard surface) external hardware such as compressors.
- Shape alloys in which the object returns to a specified shape when heated.
- Dielectric elastomer actuators (electrostatics) in which current is applied to change the shape of the object.
- Bi-hybrid muscle thin-film actuators with muscle microtissue on a surface.
- Artificial skin. Stretchable electronics on a surface which is soft yet sensitive to the environment.
- Metal electrical circuits embed in a fabric.
- Elastomers molded into shapes
- Liquid phase electronics made of Gallium & Indium. These can be designed to sense pressure, resistance, capacitance.
He concluded by examining some of challenges such as creating muscle consisting entirely of soft parts (and eliminating the need for compressors and other external supporting hardware).
Software design and systems modeling can also be challenging. In rigid systems, kinematics & dynamics can be modeled relatively easily. This means that standard CAD software can be used for path planning and to study feedback control.
By contrast soft systems must account for global kinematics and need to simulate non-linear elastic behavior, often considering fluid mechanics. This requires solving PDEs using finite element analysis on a mesh. Currently there is only limited software to do this. VoxCAD, developed at Cornell is one option and uses cube-like units for simulation (Minecraft?). Even with this software, systems need extensive evaluation for determine the accuracy of the simulation.