A three-year, $700,000 National Science Foundation (NSF) Robotics Initiative grant will enable two GT-AE researchers to tackle an issue that is critical to missions in Earth orbit and beyond: integrated sensing and planning under uncertain conditions.
Prof. Panagiotis Tsiotras and Assistant Prof. Evangelos Theodorou were informed that their proposal, “Information-Theoretic Trajectory Optimization for Motion Planning and Control with Applications to Space Proximity Operations” was funded by the NSF late last month.
This research seeks to develop the theory and algorithms that will enable active exploration using robust, reliable sensing and planning of a free-flying space robot near either a man-made or natural body -- a spacecraft or asteroid.
Assistant Prof. Evangelos Theodorou
The research shows enormous potential in space missions, where satellite servicing and refueling, space station resupply, the removal of space debris, and in-space vehicle inspections require a combination of robotic and human reliability.
Tsiotras and Theodorou anticipate that the same techniques will also be employed in all other problems where an intelligent agent needs to navigate autonomously in an uncertain and dynamically changing environment.
"Satellite and servicing and refueling in orbit is long-overdue. Technical challenges have delayed progress, but recently several efforts, spearheaded primarily by DARPA and NASA, have exerted a renewed interest in this area,” said Tsiotras, the PI for the project.
“DARPA's Phoenix program, for example, seeks to extend the satellite’s lifetime via robotic servicing and repair in orbit. In fact, just this week DARPA announced a request for information for a new program focusing specifically on robotic servicing of satellites in GEO and the transition of this technology to servicing telecommunication satellites in orbit. Needless to say, the commercialization potential in this area is huge, and Georgia Tech is at the forefront of developing the required technology.”
Theodorou is also excited about working on the project, which he says represents a departure from traditional space robotics.
"Traditionally, we would look at one way -- the most direct way -- of getting from A to B, but this is under the assumption that we operate in a deterministic world", said Theodorou. " With this project we look at different ways to get from A to B, that take into account sensor ambiguity and uncertainty"
In more technical terms:
Currently, perceptual cues (especially visual information) are treated as surrogates for full-state feedback estimators.This enforces an artificial separation between perception and control action. This dichotomy between sensory data acquisition/processing, and control/actuation strategies is unsuitable for this problem, where information gathering (perception/sensing) is tightly coupled with motion (control). To overcome the aforementioned limitations, in this work it is proposed to use tools from stochastic optimal control in order to extract actionable information from raw sensory inputs. A key ingredient of the proposed approach is to keep track of the first and second order statistics of the estimation error so that control actions depend on both of them. The result is a new, computationally more efficient, methodology to maximize information gathering during the exploration phase and to optimize over distributions of trajectories during the execution phase.