Education
- Ph.D., Carnegie Mellon University (ME)
- M.S., University of Pennsylvania (ME)
- M.S., University of Pennsylvania (EE)
- B.S., Drexel University (EE and BME)
Background
Dr. Frank L. Hammond III joined George W. Woodruff School of Mechanical Engineering in April 2015. Prior to this appointment, he was a postdoctoral research affiliate and instructor in the Department of Mechanical Engineering at MIT and a Ford postdoctoral research fellow at the Harvard School of Engineering and Applied Sciences. He received his PhD in 2010 from Carnegie Mellon University.
Research
Dr. Hammond’s research focuses on the design and control of adaptive robotic manipulation (ARM) systems. This class of devices exemplified by kinematic structures, actuation topologies, and sensing and control strategies that make them particularly well-suited to operating in unstructured, dynamically varying environments - specifically those involving cooperative interactions with humans. The ARM device design process uses an amalgamation of bioinspiration, computational modeling and optimization, and advanced rapid prototyping techniques to generate manipulation solutions which are functionally robust and versatile, but which may take completely non-biomorphic (xenomorphic) forms. This design process removes human intuition from the design loop and, instead, leverages computational methods to map salient characteristics of biological manipulation and perception onto a vast robotics design space. Areas of interest for ARM research include kinematically redundant industrial manipulation, wearable robotic devices for human augmentation, haptic-enabled teleoperative robotic microsurgery, and autonomous soft robotic platforms.
A key scientific challenge in ARM research is synthesizing robot designs which promote the functional versatility, efficiency, and mechanical robustness seen in biological manipulators, but which are built from non-biomorphic mechanisms, actuators, and energy sources. The first step in that process is empirical characterization of the biological manipulation systems that robotic systems will emulate. In the case of adaptive robotic grasping, the biomechanics and kinetics of human grasping are measured using soft wearable sensor suites (built in-lab) and various motion tracking systems. The experimental data is then analyzed using principal component analysis, partial least squares (regression), and other dimensionality reduction methods to elucidate form-function relationships and quantitative descriptions of human grasp mechanics. This information then forms the basis for the functional requirements of a robotic manipulator.
Mechanistic and statistical models generated from experimental data can be used to describe the characterized manipulation tasks mathematically. Computational models of candidate robotic manipulators – consisting of motion transmission mechanisms, actuators, and kinematic topologies, and control laws – are then used to simulate the tasks and assess manipulator design quality. The manipulator design space can include underactuated mechanisms, passively compliant structures, and distributed sensors for autonomous control or teleoperation. Various environmental conditions and task disturbances can be imposed on a manipulator in simulation, and computational design refinement can continue until certain performance criteria and design constraints are satisfied. To design kinematically-redundant industrial manipulators, for example, multiple task variants and environmental obstacles can be introduced in simulation to force design solutions which are both disturbance-tolerant and dexterous.
After optimization in-silica, ARM device designs are prototyped and experimentally validated in target environments. Rapid prototyping methods including, 3D printing, shape deposition manufacturing, and soft lithography, allow economical manufacturing of strain sensors, pneumatic actuators and other tunable, modular robotic components. Efficient, inexpensive prototyping capabilities are particularly for wearable devices which, along with mechanical performance requirements, must satisfy a variety human factors requirements. Experimental data gathered using these prototypes can be used to adjust simulation parameters and seed further optimizations.
Dr. Hammond is excited to collaborate with Georgia Tech faculty and students on a variety of ARM research projects, including new topics such as xenomorphic robotic systems, human-robot interaction and co-adaptation models, and transformable, autonomous robotic manipulation platforms.