MAE 298 Seminar:Ellipse Synthesis and the Computational Design of Robot Geometries

Abstract: The field of robotics is shaped by the tools used to solve its underlying problems. Despite 
increasing sophisticated computational advances in control, perception, and planning, the area of mechanical 
design still depends on strokes of genius from clever individuals to dream up new robot mechanisms and 
morphologies that move the field forward. The goal of my research is to replace intuition with computation in 
the geometric design of robots. This presentation will describe how to mathematically frame the design of 
multi-DOF manipulators. Jacobian ellipses are used to specify desired multi-directional transmission 
characteristics from actuators to end-effector. The forward problem, manipulator-to-ellipses, is classical 
theory. The inverse problem, ellipses-to-manipulator, is a design problem and an open challenge. As a 
nonlinear problem, it possesses many minima across a diverse design space. Local optimizers need a good 
guess, making them suitable for design tuning but not design space exploration. To conduct the latter, tandem 
neural networks and monodromy continuation afford two appropriate pathways forward. The framework and 
solution techniques of ellipse synthesis enable the multi-directional customization of force, velocity, stiffness, 
backdrivability, precision, sensitivity, and power distribution properties across the workspace of a manipulator. 
Traditional trade-offs between these quantities are overcome through reconfigurability without requiring extra 
actuators. Discussed examples include legged robots, grippers, biomimicry, and deployable aircraft.
Bio: Mark Plecnik is an Associate Professor in the Department of Mechanical Engineering at the University of 
Utah. His research spans from computational & theoretical activities to prototyping & experimentation, with 
all contributions grounded in the computational design of geometry for new robotic inventions. He is currently 
focusing on forming configuration surfaces to influence prevailing dynamics, using neural nets to synthesize 
manipulators with customized multi-directional transmission characteristics, and integrating reconfiguration into 
design tools for overcoming traditional trade-offs. Plecnik received the NSF CAREER Award in 2022. He 
received the Freudenstein Young Investigator Award from ASME in 2024. He is a senior member of IEEE