Electric vertical takeoff and landing vehicles are becoming promising for on-demand air transportation in urban mobility (UAM). However, successfully bringing such airspace operations to fruition will require introducing orders of magnitude more aircraft a given volume. Although there existing solutions communication technology, onboard computing capability, sensor the computation guidance algorithm enable safe, efficient, scalable flight dense self-organizing traffic still remains an open...

The use of electrical vertical takeoff and landing (eVTOL) aircraft to provide efficient, high-speed, on-demand air transportation within a metropolitan area is topic increasing interest, which expected bring fundamental changes the city infrastructures daily commutes. NASA, Uber, Airbus have been exploring this exciting concept Urban Air Mobility (UAM), has potential meaningful door-to-door trip time savings compared with automobiles. However, ability manage many these eVTOL safely in...

Obstacle avoidance for small unmanned aircraft is vital the safety of future urban air mobility (UAM) and Unmanned Aircraft System (UAS) Traffic Management (UTM). There are a variety techniques real-time robust drone guidance, but numerous them solve in discretized airspace control, which would require an additional path smoothing step to provide flexible commands UAS. To deliver safe computationally efficient guidance UAS operations, we explore use deep reinforcement learning algorithm...

The concept of Urban Air Mobility (UAM) proposes to use revolutionary new electrical vertical takeoff and landing (eVTOL) aircraft provide efficient on-demand air transportation service between places previously underserved by the current aviation market. A key challenge for success UAM is how manage large-scale autonomous flight operations with safety guarantee in high-density, dynamic uncertain airspace environments. In this paper, a assured decentralized online guidance algorithm airborne...

Experience replay is widely used in deep reinforcement learning algorithms and allows agents to remember learn from experiences the past. In an effort more efficiently, researchers proposed prioritized experience (PER) which samples important transitions frequently. this paper, we propose Prioritized Sequence Replay (PSER) a framework for prioritizing sequences of attempt both efficiently obtain better performance. We compare performance PER PSER sampling techniques tabular Q-learning...

A novel deep multi-agent reinforcement learning framework is proposed to identify and resolve conflicts among a variable number of aircraft in high-density, stochastic, dynamic sector. Currently the sector capacity constrained by human air traffic controller's cognitive limitation. We investigate feasibility new concept (autonomous separation assurance) approach push above propose using distributed vehicle autonomy ensure separation, instead centralized controller. Our utilizes Proximal...

A novel deep multi-agent reinforcement learning framework is proposed to identify and resolve conflicts among a variable number of aircraft in high-density, stochastic, dynamic en route sector. The concept using distributed vehicle autonomy ensure separation proposed, instead centralized sector air traffic controller. Our uses proximal policy optimization that customized incorporate an attention network. This allows the agents have access information scalable, efficient approach achieve high...

Urban Air Mobility (UAM) is an emerging transportation mode, where electrical vertical takeoff and landing (eVTOL) aircraft will transport cargo passengers within a city. In our previous work, we have introduced computational guidance algorithms for single multiple cooperative to navigate through obstacles avoid conflicts among aircraft. However, assumed perfect communications air-to-air ground-to-air channels. this paper, formulate the problem with communication constraints loss....

The emergence of new operations, such as package delivery and air taxis, that could co-exist in the current airspace has been a topic great interest recent years. A big challenge associated with introduction these operations is ensuring complex system-of-systems responsible for known Unmanned Traffic Management (UTM) meet high safety standards aviation. In this work, we present general purpose framework validating decision making systems, protocols, algorithms may exist UTM ecosystem. We...

Ensuring safety and providing obstacle conflict alerts to small unmanned aircraft is vital their integration into civil airspace. There are many techniques for real-time robust drone guidance, but of them need expensive computation time or large memory requirements, which not applicable deploy onboard an with limited resources. To provide a safe efficient computational guidance operations aircraft, we framework using deep reinforcement learning algorithm guide autonomous UAS destinations...

Computer vision is a key component for safe autonomous flying of small UAS. For example, in package delivery mission, or an emergency landing event, pedestrian detection could help UASs with zone identification. In this research, we focus on deep-learning-based computer UAS and tracking. contrast existing research ground-level detection, our contribution that achieve highly accurate multiple from birds-eye view, when both the pedestrians platform are moving.

Decision-making agents with planning capabilities have achieved huge success in the challenging domain like Chess, Shogi, and Go. In an effort to generalize ability more general tasks where environment dynamics are not available agent, researchers proposed MuZero algorithm that can learn dynamical model through interactions environment. this paper, we provide a way necessary theoretical results extend generalized environments continuous action space. Through numerical on two relatively...

Markov Decision Processes (MDPs) are a mathematical framework for modeling sequential decision making under uncertainty. The classical approaches solving MDPs well known and have been widely studied, some of which rely on approximation techniques to solve with large state space and/or action space. However, most these solution their still take much computation time converge usually must be re-computed if the reward function is changed. This paper introduces novel alternative approach exactly...

An approach is given, which tries to design a robust controller using the plicy iteration algorithm for nonlinear systems due uncertainties. The problem divided into two steps. first step introducting uncertainties cost function changes difficult point optimal problem. Second, obtain control strategy optimization of new cosst function, policy technique utilized deal with Hamilton-Jacobi-Bellman (HJB) equation. In implementation given approach, obtained approximately by constructing critic...

Obstacle avoidance for small unmanned aircraft is vital the safety of future urban air mobility (UAM) and Unmanned Aircraft System (UAS) Traffic Management (UTM). There are many techniques real-time robust drone guidance, but them solve in discretized airspace control, which would require an additional path smoothing step to provide flexible commands UAS. To a safe efficient computational guidance operations aircraft, we explore use deep reinforcement learning algorithm based on Proximal...