Erik Komendera

Erik Komendera

Curriculum Vitae

I am inspired by the ways that living creatures can achieve complex goals such as assembly and movement, and I want to explore ways to model and improve these methods.  I have focused this inspiration into two main areas: intelligent assembly and nonlinear control, both of which are manifest in nature: assembly methods ranging from termite mound construction to highly organized human construction projects; and the highly adaptable movement methods embodied in all varieties of animals. These subjects are both hot topics in robotics research, but currently robots cannot do either very well.  The assembly interest has led to a NASA Space Technology Research Fellowship to explore orbital assembly, while the movement interest has filled out the remainder of my time and will likely be the subject of my PhD thesis.  Specifically, my current research projects are:

Intelligent Precision Jigging: a system for assembling and welding precise trusses on orbit. I am doing this in conjunction with John Dorsey and William Doggett at the NASA Langley Research Center.  Our long-term goals are to develop a system that can be stationed on orbit and can assemble structures given the materials.  This system can be reusable, making it ideal for an automated construction station. The requirement for precise jigging is embedded in the assembly robots, allowing the use of simple node balls and extensible struts. My role is to develop the Intelligent Precise Jigging Robot (IPJR), a device that can position itself precisely on the location of the next truss cell, and hold the nodes and struts while an auxiliary arm welds. The auxiliary arm will then reposition the IPJR to the location of the next truss cell, and this process repeats. This research is paid for by the NASA Space Technology Research Fellowship.

An early IPJR prototype, developed during my Visiting Fellowship at Harvard:

Demonstration of the intelligent scaffolding assembly algorithm, in which assembly robots navigate the exterior of an in-progress assembly, and place blocks in an efficient manner:

Tensegrity Robots: robots embodied as tensegrity structures may hold several benefits, including:

  1. Compliance: safe in the neighborhood of fragile objects (such as humans).
  2. Deformability: suitable for highly constricted environments.
  3. Tolerance to arbitrary high loads: resistant to unexpected forces due to efficient redistribution of loads.
  4. High degrees of freedom: robust to motion requirements in unstructured environments.
  5. High stiffness-to-mass ratio: light yet strong.
  6. Failure tolerance: still operable despite failures in individual components.
  7. Easy to model: only axial forces are considered, allowing the use of ODE solvers instead of PDE solvers.

Tensegrity structures can be found in nature, from individual cell structure to the musculoskeletal system of animals.  My interest in tensegrity robotics is in intelligent control and path planning, in both slowly varying and highly dynamics systems.  Unlike conventional robotics with rigid joints, inverse kinematics and dynamics cannot easily be calculated, which has hindered their usefulness to date.

Demonstration of my tensegrity robot simulator:

Soft Robots: Many of the advantages and challenges in tensegrity robotics can be applied to soft robotics, and my interests in soft robots also lie in intelligent control and path planning.

Chaotic System Reachability Set Exploration: My interest in searching and controlling the dynamical state space of tensegrity and soft robots extends to general chaotic systems.  To this end, I have created a new search algorithm to efficiently map the reachability sets of spacecraft n-body problems.  These maps can be used for path planning and control of spacecraft trajectories in highly chaotic systems, and also can be used by human planners to better understand the dynamics of such systems.  This work has led to two conference publications, and a journal article is in progress.

Visualization of chaotic spacecraft trajectories in a restricted three-body problem: