We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

eess.SY

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Electrical Engineering and Systems Science > Systems and Control

Title: Collision-Free Flocking with a Dynamic Squad of Fixed-Wing UAVs Using Deep Reinforcement Learning

Abstract: Developing the collision-free flocking behavior for a dynamic squad of fixed-wing UAVs is still a challenge due to kinematic complexity and environmental uncertainty. In this paper, we deal with the decentralized leader-follower flocking control problem through deep reinforcement learning (DRL). Specifically, we formulate a decentralized DRL-based decision making framework from the perspective of every follower, where a collision avoidance mechanism is integrated into the flocking controller. Then, we propose a novel reinforcement learning algorithm CACER-II for training a shared control policy for all the followers. Besides, we design a plug-n-play embedding module based on convolutional neural networks and the attention mechanism. As a result, the variable-length system state can be encoded into a fixed-length embedding vector, which makes the learned DRL policies independent with the number or the order of followers. Finally, numerical simulation results demonstrate the effectiveness of the proposed method, and the learned policies can be directly transferred to semiphysical simulation without any parameter finetuning.
Comments: 8 pages, 8 figures
Subjects: Systems and Control (eess.SY); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
Cite as: arXiv:2101.08074 [eess.SY]
  (or arXiv:2101.08074v1 [eess.SY] for this version)

Submission history

From: Chao Yan [view email]
[v1] Wed, 20 Jan 2021 11:23:35 GMT (807kb,D)

Link back to: arXiv, form interface, contact.