CS Events Monthly View

Faculty Candidate Talk

Robotic Planning Through the Lens of Optimality: Blessings in a Curse

 

Download as iCal file

Thursday, March 12, 2015, 10:30am

 

Arguably, planning plays the most central role in robotics research: all robots must continuously sense, compute a plan, and execute the plan. Frequently, it is also highly desirable to carry out the planning phase optimally, which directly translates to minimizing fuel consumption, task completion time, and so on. However, aiming for optimal solutions often greatly complicates robotic planning problems. In this talk, I will touch upon on three such problems in two application domains: multi-robot coordination/control and optimal tour planning for a mobile sensing robot. I will show that, whereas the introduction of optimality requirements readily renders such problems more difficult (sometimes NP-hard), the additional structure induced by the requirements can in fact help us significantly. For each of the three problems, the additional structure (distinct in each case) either leads to polynomial-time optimal algorithms or much more efficient computation (e.g., 100x speedups) of difficult problem instances.

Speaker: Jingjin Yu

Bio

Jingjin Yu is a postdoctoral researcher in the Computer Science and Artificial Intelligence Lab (CSAIL) at the Massachusetts Institute of Technology. He obtained his PhD degree in Electrical and Computer Engineering, and his MS degree in Computer Science,

Location : CoRE Lecture Hall (Room 101)

Committee

Dimitris Metaxas

Event Type: Faculty Candidate Talk

Abstract: 

Organization

Massachusetts Institute of Technology