Memory-enhanced deep reinforcement learning for UAV navigation in 3D environment

Article Computer Science, Artificial Intelligence

Tactical UAV path optimization under radar threat using deep reinforcement learning

M. Nedim Alpdemir

Summary: This paper proposes a RL based solution to the problem of tactical UAV path planning under hostile radar tracking threat. The solution integrates a Markov Decision Process (MDP) compliant RL environment with comprehensive probabilistic radar behavior models, and combines a core RL algorithm (DQN with Prioritized Experience Replay) with a transfer learning variant (Learning from Demonstrations) to improve the performance of a learning agent and address the issue of sparse rewards.

NEURAL COMPUTING & APPLICATIONS (2022)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Exploiting a fleet of UAVs for monitoring and data acquisition of a distributed sensor network

Somaiyeh MahmoudZadeh et al.

Summary: This study introduces an efficient data collection strategy utilizing a team of UAVs to monitor and collect data from a large distributed sensor network. The collaborative mission planning system developed enables UAVs to collectively conduct data collection missions, maximizing the number of visited sensor nodes through optimization algorithms. Simulation results confirm the effectiveness and reliability of the proposed mission planner for distributed sensor network monitoring and data collection.

NEURAL COMPUTING & APPLICATIONS (2022)

添加到收藏夹

Article Computer Science, Artificial Intelligence

UAV object tracking by background cues and aberrances response suppression mechanism

Tian Li et al.

Summary: Real-time object tracking for UAVs is crucial in computer vision, but challenging due to complex scenarios and small targets. This paper proposes a method utilizing background cues and aberrances response suppression mechanism for 4 degrees-of-freedom tracking in UAV scenes. By decomposing the tracking task into subproblems and introducing a logarithmic polar coordinate system, accurate target states are obtained through phase correlation calculations and response fusion. Experiments show superior performance in small target tracking compared to current advanced trackers.

NEURAL COMPUTING & APPLICATIONS (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Neural network and fuzzy logic-based hybrid attitude controller designs of a fixed-wing UAV

Saban Ulus et al.

Summary: This study evaluated various control techniques and selected the best controller structure for small UAV control. The lateral dynamics were successfully simulated, and five different controller methods were used for simulations, with results showing that the combined PD-fuzzy-PI and ANFIS-PID controllers performed better than others under steady flight conditions.

NEURAL COMPUTING & APPLICATIONS (2021)

添加到收藏夹

Article Engineering, Civil

Data Freshness and Energy-Efficient UAV Navigation Optimization: A Deep Reinforcement Learning Approach

Sarder Fakhrul Abedin et al.

Summary: This study presents a navigation policy for unmanned aerial vehicles and designs an energy-efficient trajectory optimization problem with a deep reinforcement learning model to improve energy efficiency and data freshness. The proposed approach outperforms greedy and baseline methods by 3.6% and 3.13% in terms of energy efficiency according to simulation results.

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence