Minimality and Trade-offs in Automated Robot Design


Organizers: Alexandra Nilles, Hadas Kress-Gazit, Jason O'Kane, Andrea Censi

Website: http://minimality.mit.edu/

Most fields of engineering are characterized by fundamental trade-offs between maximizing performance and minimizing resource usage; robotics is no exception. In robot design, trade-offs are distributed among subsystems such as sensing, actuation, computation, and power. A mathematical framework to describe these trade-offs does not yet exist.

The speakers and discussions in this workshop will focus on working toward formal representations that make automated reasoning and synthesis possible, and enable design choices beyond once-off, ad hoc solutions.

The workshop brings together roboticists, designers and makers with a variety of backgrounds to start to answer the question: "How can computers and software help us navigate the space of design decisions?"

The workshop welcomes contributions about:

  1. practical examples of resource-constrained robots and their performance envelopes;
  2. abstractions and models which have promise as foundations for algorithmic design;
  3. novel and unconventional ideas for how to tame the computational complexities involved.

The intended outcome of the workshop is a better understanding of how informal and intuitive design decisions, including those exemplified in a custom design card game — played and discussed after lunch — can be refined and systematized so as to become fit for automation. The organizers will coordinate with the speakers a position paper that synthesizes the points of view described in the workshop. Target journals could be: Robotics & Automation Letters; Nature Robotics; and Robotics & Automation Magazine.