TY - GEN
T1 - Learning distributed controllers for V-formation
AU - Roy, Shouvik
AU - Mehmood, Usama
AU - Grosu, Radu
AU - Smolka, Scott A.
AU - Stoller, Scott D.
AU - Tiwari, Ashish
N1 - Publisher Copyright: © 2020 IEEE.
PY - 2020/8
Y1 - 2020/8
N2 - We show how a high-performing, fully distributed and symmetric neural V-formation controller can be synthesized from a Centralized MPC (Model Predictive Control) controller using Deep Learning. This result is significant as we also establish that under very reasonable conditions, it is impossible to achieve V-formation using a deterministic, distributed, and symmetric controller. The learning process we use for the neural V-formation controller is significantly enhanced by CEGkR, a CounterexampleGuided k-fold Retraining technique we introduce, which extends prior work in this direction in important ways. Our experimental results show that our neural V-formation controller generalizes to a significantly larger number of agents than for which it was trained (from 7 to 15), and exhibits substantial speedup over the MPC-based controller. We use a form of statistical model checking to compute confidence intervals for our neural Vformation controller's convergence rate and time to convergence.
AB - We show how a high-performing, fully distributed and symmetric neural V-formation controller can be synthesized from a Centralized MPC (Model Predictive Control) controller using Deep Learning. This result is significant as we also establish that under very reasonable conditions, it is impossible to achieve V-formation using a deterministic, distributed, and symmetric controller. The learning process we use for the neural V-formation controller is significantly enhanced by CEGkR, a CounterexampleGuided k-fold Retraining technique we introduce, which extends prior work in this direction in important ways. Our experimental results show that our neural V-formation controller generalizes to a significantly larger number of agents than for which it was trained (from 7 to 15), and exhibits substantial speedup over the MPC-based controller. We use a form of statistical model checking to compute confidence intervals for our neural Vformation controller's convergence rate and time to convergence.
KW - Distributed Neural Controller
KW - Model Predictive Control
KW - Supervised Learning.
KW - V-Formation
UR - https://www.scopus.com/pages/publications/85092711567
U2 - 10.1109/ACSOS49614.2020.00033
DO - 10.1109/ACSOS49614.2020.00033
M3 - Conference contribution
T3 - Proceedings - 2020 IEEE International Conference on Autonomic Computing and Self-Organizing Systems, ACSOS 2020
SP - 119
EP - 128
BT - Proceedings - 2020 IEEE International Conference on Autonomic Computing and Self-Organizing Systems, ACSOS 2020
A2 - El-Araby, Esam
A2 - Tomforde, Sven
A2 - Wood, Timothy
A2 - Kumar, Pradeep
A2 - Raibulet, Claudia
A2 - Petri, Ioan
A2 - Valentini, Gabriele
A2 - Nelson, Phyllis
A2 - Porter, Barry
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 1st IEEE International Conference on Autonomic Computing and Self-Organizing Systems, ACSOS 2020
Y2 - 17 August 2020 through 21 August 2020
ER -