MIT AeroAstro — Lincoln Laboratory Undergraduate Research and Innovation Scholar
Scheduling and Controllability for Temporal Constraint Networks with Uncontrollable Loops
Brian C. Williams
Scientists and Autonomous Underwater Vehicles operators routinely use actions that involve “looping behavior,” i.e. actions that consist of an integer number of repeated sub-actions, often with uncontrollable duration. This is commonly seen in a coverage pattern consisting of repeated passes back and forth over a region of interest in the shape of a grid. This pattern is also seen in UAV search and rescue missions, manufacturing robots repeatedly picking and placing components, etc. During this SuperUROP, I plan to develop temporal reasoning algorithms to determine the feasibility of temporal networks that contain uncertain looping durations. If a temporal constraint network is controllable, that implies that the plan can be executed regardless of the uncertainty introduced by nature.
I am especially interested in autonomous research like this because autonomous systems are crucial for space exploration. One of my career goals is to help develop fully autonomous systems enabling exploration of extraterrestrial environments and deep space. To prepare myself, I am taking 16.410/413 this semester, which focuses on planning and constraint programming algorithms similar to what I will be designing.