Feb. 17, 11:45, Wei Ji Ma. Optimality and Probabilistic Computation in Visual Categorization

Kostas Bekris

Department of Computer Science and CBIM, Rutgers

Progress in algorithmic motion planning and opportunities at the intersection with perceptual science

The area of motion planning studies algorithms that return actions

that must be executed to accomplish a physical task. It has

applications in robotics, simulation, cyber-physical systems and

human-centered computing. This talk will first review recent progress

in the field, specifically regarding (a) the efficient computation of

asymptotically (near-)optimal paths, (b) dealing with systems that

exhibit non-trivial dynamics, (c) and providing efficient algorithms

for multi-agent path-finding. It will then progress to view these

contributions under the light of recent efforts, such as the National

Robotic Initiative, which emphasize the importance of bringing robots

and people together to solve complex tasks. The talk will propose a

framework for reasoning about such interactions in an algorithmic

manner, utilizing ideas from game theory. This direction provides

research opportunities that lie at the intersection of perceptual

science and motion planning, such as utilizing and analyzing

perceptual data in order to identify the intent of humans by robots

that will assist in the completion of a common task.

Background readings:

Books on motion planning:

    - "Principles of Robot Motion, Theory, Algorithms, and Implementations" by Howie Choset, Kevin M. Lynch, Seth Hutchinson, George Kantor, Wolfram Burgard, Lydia E. Kavraki and Sebastian Thrun

    - "Planning Algorithms" (/open-access book/) by Steve LaValle

Related Papers: