ERI: Scalable Distributed Optimal Control Algorithms Validated On Multi-Robot Test-bed

This Engineering Research Initiation (ERI) award supports research that looks to solve the coupled problem of optimal and stable topology selection for multi-robot systems in a real-world environment, thereby promoting the progress of science and securing the national defense. This project will focus on development of distributed solutions to the optimal topology selection problem among robots with asymmetric interactions where robots have different sensor modalities. Conventionally, centralized computation and perfect localization is assumed for multi-robot systems. However, when a system is deprived of such a setting, robots in a team must rely on robust one-to-one interaction using on-board sensors. This project looks to select an optimal set of robot-robot interactions to conduct stable multi-robot coordination using distributed algorithms. This research seeks to enable more robust coordination of multi-robot systems to environmental conditions such that these teams of robots can be used in out of lab settings without depending on any centralized base station. Multi-robot coordination has viable applications from enhancing agricultural yield by monitoring crops, to automating safety of critical infrastructure like power stations, and acting against adversaries. Furthermore, both undergraduate and graduate students will have the opportunity to participate in interdisciplinary working groups for early career professionals focused on translational research and pedagogy.

This research aims to make fundamental contributions to multi-robot systems which are stable and robust in coordination to carry out tasks. It looks to achieve this goal by formulating an optimal topology selection problem and developing scalable control-theoretic methods for topology control that guarantee stability and robustness to disturbance in the presence of asymmetric robot interactions. The project takes a multi-tiered approach to: i) solve a distributed optimal topology selection problem; ii) validate the algorithms in high-fidelity physics based simulations, and iii) test the algorithms on a team of unmanned aerial vehicles, in real-world conditions using the off-the-shelf robotic and sensor platforms. The optimal topology selection problem for multi-robot systems will be formulated as complex mixed integer semidefinite programs. This research seeks to extend the scope of applying chordal decomposition combined with the alternating direction method of multipliers method to exploit sparsity in large semidefinite programs. The testbed will be developed at the state-of-the-art facility at Florida Atlantic University’s SeaTech, Dania Beach campus which is already equipped with a team of unmanned aerial vehicles and a OptiTrack motion capture system for robot localization.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Award Number: 2503594
Principal Investigator: Pratik Mukherjee
Funds Obligated: $200,000
State: FL