This paper presents the development and hardware implementation of an autonomous battery maintenance mechatronic system that significantly extends operational time powered small-scaled unmanned aerial vehicles (UAVs). A simultaneous change charge approach is used to overcome significant downtime experienced by existing charge-only approaches. The automated quickly swaps a depleted UAV with replenished one while simultaneously recharging several other batteries. results in low downtime,...

Markov decision processes (MDPs) are often used to model sequential problems involving uncertainty under the assumption of centralized control. However, many large, distributed systems do not permit control due communication limitations (such as cost, latency or corruption). This paper surveys recent work on decentralized MDPs in which each agent depends a partial view world. We focus general framework where there may be about state environment, represented partially observable MDP...

Autonomous lane changing is a critical feature for advanced autonomous driving systems, that involves several challenges such as uncertainty in other driver's behaviors and the trade-off between safety agility. In this work, we develop novel simulation environment emulates these train deep reinforcement learning agent yields consistent performance variety of dynamic uncertain traffic scenarios. Results show proposed data-driven approach performs significantly better noisy environments...

This paper presents the design, analysis and experimental testing of a variablepitch quadrotor. A custom in-lab built quadrotor with on-board attitude stabilization is developed tested. An dynamic differences in thrust output between fixed-pitch variable-pitch propeller given validated simulation results. It shown that actuation has significant advantages over conventional configuration, including increased rate change, decreased control saturation, ability to quickly efficiently reverse...

Gas turbine maintenance requires consistent inspections of cracks and other structural anomalies. The provide information regarding the overall condition structures yield for estimating health repair costs. Various image processing techniques have been used in past to address problem automated visual crack detection with varying degrees success. In this work, we propose a novel framework that utilizes from both classical deep learning methodologies. main contribution work is demonstrating...

In persistent missions, taking system's health and capability degradation into account is an essential factor to predict avoid failures. The state space in health-aware planning problems often a mixture of continuous vehicle-level discrete mission-level states. This particular poses challenge when the mission domain partially observable restricts use computationally expensive forward search methods. paper presents method that exploits structure exists many performs two-layer scheme. lower...

There is a growing demand for unmanned air vehicles (UAVs) with combat capabilities in battlefield scenarios [1]. Whether this capability evasive maneuvers or flying attack patterns, (UCAVs) are expected to operate dense and often threatening environments that require aggressive trajectory planning controls These trajectories the use of maneuvering over full flight envelope aircraft. Examples such high-g turns high angle-of-attack maneuvers. This article presents development multimodal...

This video submission presents a design concept of an autonomous variable-pitch quadrotor with constant motor speed. The main aim this work is to increase the maneuverability vehicle while largely maintaining its mechanical simplicity. added will allow agile maneuvers like inverted hover and flip. A custom in lab built onboard attitude stabilization developed tested ACL's (Aerospace Controls Laboratory) RAVEN (Real-time indoor Autonomous Vehicle test ENvironment). Initial flight results show...

In this paper, we present an advanced adaptive cruise control (ACC) concept powered by Deep Reinforcement Learning (DRL) that generates safe, human-like, and comfortable car-following policies. Unlike the current trend in developing DRL-based ACC systems, propose defining action space of DRL agent with discrete actions rather than continuous ones, since human drivers never set throttle/brake pedal level to be actuated, but required change levels. Through human-like throttle-brake...

The popularity of commercial unmanned aerial vehicles has drawn great attention from the e-commerce industry due to their suitability for last-mile delivery. However, organization multiple efficiently delivery within limitations and uncertainties is still a problem. main challenge planning scalability, since space grows exponentially number agents, it not efficient let human-level supervisors structure problem large-scale settings. Algorithms based on Deep Q-Networks had unprecedented...

In this paper we study the localization and tracking of a radio frequency (RF) emitting target using multiple unmanned aerial vehicles (UAVs) over large scale environment. Although RF targets measurements is well studied problem, standard approaches become inefficient when signal power uncertain there significant noise in received strength (RSS) search environment scale. We present architecture, where data driven neural network model used for estimating unknown extended Kalman filters are...

In this paper, we investigate the effect of local disturbances on European airports over global delay characteristics air traffic network. First, existing data is used for analyzing busiest and their connectivity to other in Based analysis, an airport based queuing network model constructed simulating propagation across The generating various scenarios where capacities were reduced under (weather effects, controller strikes etc.). consequences these capacity reductions total (departure +...

Abstract Navigation and planning for unmanned aerial vehicles (UAVs) based on visual-inertial sensors has been a popular research area in recent years. However, most visual are prone to high error rates when exposed disturbances such as excessive brightness blur, which can lead catastrophic performance drops perception motion systems. This study proposes novel framework address the coupled perception-planning problem high-risk environments. achieved by developing algorithms that...

In large-scale persistent missions, the vehicle capabilities and health often degrade over time. This paper presents a Health Aware Planning (HAP) Framework for long-duration complex UAV missions by establishing close feedback between high-level planning based on Markov Decision Processes (MDP) execution level learning-focused adaptive controllers. enables HAP framework to plan anticipating failures reassessing after failures. proactive behavior allows efficient replanning account changing...

Planning, control, perception, and learning are current research challenges in multirobot systems. The transition dynamics of the robots may be unknown or stochastic, making it difficult to select best action each robot must take at a given time. observation model, function robots' sensor systems, noisy partial, meaning that deterministic knowledge team's state is often impossible attain. Moreover, actions can have an associated success rate and/or probabilistic completion Robots designed...

Commonly used Proportional-Integral-Derivative based UAV flight controllers are often seen to provide adequate trajectory-tracking performance only after extensive tuning.The gains of these tuned particular platforms, which makes transferring from one other time-intensive.This paper suggests the use adaptive in speeding up process extracting good control new UAVs.In particular, it is shown that a concurrent learning controller improves trajectory tracking quadrotor with baseline linear...

This paper introduces a novel hierarchical decomposition approach for solving Multiagent Markov Decision Processes (MMDPs) by exploiting coupling relationships in the reward function. MMDP is natural framework stochastic multi-stage multiagent decision-making problems, such as optimizing mission performance of Unmanned Aerial Vehicles (UAVs) with health dynamics. However, computing optimal solutions often intractable because state-action spaces scale exponentially number agents. Approximate...

Automated lane change is one of the most challenging task to be solved highly automated vehicles due its safety-critical, uncertain and multi-agent nature. This paper presents novel deployment state art Q learning method, namely Rainbow DQN, that uses a new safety driven rewarding scheme tackle issues in an dynamic simulation environment. We present various comparative results show our approach having reward feedback from layer dramatically increases both agent's performance sample...

In recent years, both the scientific community and industry have focused on moving computational resources with remote data centres from centralized cloud to decentralised computing, making them closer source or so called "edge" of network. This is due fact that system alone cannot sufficiently support huge demands future networks massive growth new, time-critical applications such as self-driving vehicles, Augmented Reality/Virtual Reality techniques, advanced robotics critical control...