1 of 91

Enhancing Safety and Energy-Efficiency in Advanced Air Mobility Through Trajectory Planning and Mission Feasibility Assessment Strategies

PhD Defense | July 1, 2025

Abenezer Girma Taye

2 of 91

Enhancing Safety and Energy-Efficiency in Advanced Air Mobility Through Trajectory Planning and Mission Feasibility Assessment Strategies

2

3 of 91

Enhancing Safety and Energy-Efficiency in Advanced Air Mobility Through Trajectory Planning and Mission Feasibility Assessment Strategies

Fatalities

3

Challenges with the current ground-based transportation

In 2023, 40,990 people lost their lives in motor vehicle crashes in the United States [1]

[1] National Highway Traffic Safety Administration.

4 of 91

Enhancing Safety and Energy-Efficiency in Advanced Air Mobility Through Trajectory Planning and Mission Feasibility Assessment Strategies

Fatalities

Traffic Congestion

4

Challenges with the current ground-based transportation

In 2024, average U.S. driver lost 43 hours to traffic congestion [2]

[2] INRIX Global Traffic Scorecard

5 of 91

Enhancing Safety and Energy-Efficiency in Advanced Air Mobility Through Trajectory Planning and Mission Feasibility Assessment Strategies

Fatalities

Traffic Congestion

5

Challenges with the current ground-based transportation

The largest contributor to U.S. greenhouse gas emissions 29% [3]

[3] U.S. Environmental Protection Agency

Emission

6 of 91

Enhancing Safety and Energy-Efficiency in Advanced Air Mobility Through Trajectory Planning and Mission Feasibility Assessment Strategies

Fatalities

Traffic Congestion

Expensive

6

Challenges with the current ground-based transportation

The second-largest expense in a typical American household [4]

[4] Bureau of Transportation Statistics

Emission

7 of 91

Enhancing Safety and Energy-Efficiency in Advanced Air Mobility Through Trajectory Planning and Mission Feasibility Assessment Strategies

Fatalities

Traffic Congestion

Expensive

Stressful

7

Challenges with the current ground-based transportation

Higher risk of depression and anxiety － elevate cardiovascular health risks [5]

[5] Yuliya Y Shitova. The impact of long-distance travel to work on the health of commuting labour migrants: a literature review. Population and Economics, 2024.

Emission

8 of 91

Advance Air Mobility (AAM)

8

Image credit – airdatanews, Getty/PhonlamaiPhoto, Unmanned Airspace

What is AAM? “An emerging sector in the aerospace industry that aims to safely and efficiently integrate highly automated aircraft into the national airspace system (NAS)”

[6] https://www.faa.gov/air_traffic/publications/atpubs/aim_html/chap11_section_6.html

THE GEORGE WASHINGTON UNIVERSITY

9 of 91

[7] Video credit: https://www.jobyaviation.com/

10 of 91

AAM Challenges

10

[8] George Price, Douglas Helton, Kyle Jenkins, Mike Kvicala, Steve Parker, Russell Wolfe, Felix A Miranda, Kenneth H Goodrich, Min Xue, Karen Tung Cate, et al. Urban air mobility operational concept (opscon) passenger-carrying operations. 2020.

THE GEORGE WASHINGTON UNIVERSITY

11 of 91

AAM Challenges

11

Safe and Scalable Real-Time Trajectory Planner for UAM Operations

Chapter 2

Energy-Efficient Trajectory Planning and Mission Feasibility Assessment Framework

Chapter 3

Online Mission Feasibility Assessment and Contingency Management Framework

Chapter 4

[8] George Price, Douglas Helton, Kyle Jenkins, Mike Kvicala, Steve Parker, Russell Wolfe, Felix A Miranda, Kenneth H Goodrich, Min Xue, Karen Tung Cate, et al. Urban air mobility operational concept (opscon) passenger-carrying operations. 2020.

THE GEORGE WASHINGTON UNIVERSITY

12 of 91

AAM Challenges

12

Online Mission Feasibility Assessment and Contingency Management Framework

Chapter 4

Energy-Efficient Trajectory Planning and Mission Feasibility Assessment Framework

Chapter 3

Safe and Scalable Real-Time Trajectory Planner for UAM Operations

Chapter 2

LLM-based Flight Planning

[DASC, 2024]

Energy-Aware Traffic Management

[SciTech, 2025]

eVTOL Charging Infrastructure Planning

[Aviation, 2024]

Computer Vision based

Autonomous Landing

[Aviation, 2025]

[8] George Price, Douglas Helton, Kyle Jenkins, Mike Kvicala, Steve Parker, Russell Wolfe, Felix A Miranda, Kenneth H Goodrich, Min Xue, Karen Tung Cate, et al. Urban air mobility operational concept (opscon) passenger-carrying operations. 2020.

THE GEORGE WASHINGTON UNIVERSITY

13 of 91

Problem Statement Summary

13

Develop trajectory planning and mission feasibility assessment strategies to

enable safe, scalable, energy-efficient, and resilient operations in AAM.

Overarching Research Objective

THE GEORGE WASHINGTON UNIVERSITY

14 of 91

Problem Statement Summary

14

Develop trajectory planning and mission feasibility assessment strategies to

enable safe, scalable, energy-efficient, and resilient operations in AAM.

Overarching Research Objective

How can we generate energy-efficient trajectories and assess mission feasibility?

How can we assess the feasibility of a flight mission online and manage a contingency?

How can we ensure both safety and scalability

in UAM trajectory planning?

Chapter 2

Chapter 3

Chapter 4

THE GEORGE WASHINGTON UNIVERSITY

15 of 91

Safe and Scalable Trajectory Planner for

Urban Air Mobility Operations

[9]Abenezer G. Taye, Josh Bertram, Chuchu Fan, and Peng Wei. “Reachability-based online safety verification for high-density urban air mobility trajectory planning,” AIAA AVIATION 2022 Forum, 2022.

[10]Abenezer G. Taye, Roberto Valenti, Akshay Rajhans, Anastasia Mavrommati, Pieter J. Mosterman, and Peng Wei. “Safe and scalable real-time trajectory planning framework for urban air mobility,” Journal of Aerospace Information Systems (JAIS), 2024.

16 of 91

Motivation: Urban air mobility

16

As UAM progresses in levels of maturity 🡪 thousands of operations simultaneously Need for automated tools responsible for planning and operation execution (UAM Operator)

[10] Urban Air Mobility (UAM) Concept of Operations (ConOps) Version 2.0. 2020

THE GEORGE WASHINGTON UNIVERSITY

17 of 91

Motivation

17

Urban air mobility (UAM) is －

Multi-Agent System

Safety-Critical

THE GEORGE WASHINGTON UNIVERSITY

18 of 91

Motivation

18

Safety is the first design requirement

Multi-Agent System

Safety-Critical

Scalability is the second design requirement

General Idea

Safety － Reachability analysis based online safety verification scheme

Scalability － FastMDP based scalable decision-making scheme

THE GEORGE WASHINGTON UNIVERSITY

19 of 91

Motivation

19

Safety is the first design requirement

Multi-Agent System

Safety-Critical

Scalability is the second design requirement

General Idea

Safety － Reachability analysis based online safety verification scheme

Scalability － FastMDP based scalable decision-making scheme

THE GEORGE WASHINGTON UNIVERSITY

20 of 91

Contributions

20

Safe and scalable UAM

trajectory planner

Action shielding mechanism

to enhance safety

Reward shaping mechanism

to enhance safety

State

THE GEORGE WASHINGTON UNIVERSITY

21 of 91

Problem Formulation

Ownship

Intruder

Unsafe

Area

Unsafe

Area

Free-flight based urban air mobility operation

Goal

Progress to assigned vertiport － Avoid going into “Unsafe Area”

Vertiport

THE GEORGE WASHINGTON UNIVERSITY

22 of 91

Problem Formulation

Ownship

Intruder

Unsafe

Area

Unsafe

Area

Assumptions

Noise-free sensing scheme － ownship has access to current position of nearby aircraft

Homogeneous fleet of aircraft － each aircraft has the same dynamics

Vertiport

THE GEORGE WASHINGTON UNIVERSITY

23 of 91

Problem Formulation

Ownship

Intruder

Unsafe

Area

Markov Decision Process

Reachability Analysis

THE GEORGE WASHINGTON UNIVERSITY

24 of 91

MDP Formulation

24

Formulation: Markov decision process based safe and scalable trajectory planner

Urban Air Mobility Scenario

Reward Function

V

Unsafe

area

Unsafe

area

Ownship

Vertiport

Guidance Model [4]

[11] Beard, R. W., and McLain, T. W., Small unmanned aircraft: Theory and practice, Princeton University Press, 2012

THE GEORGE WASHINGTON UNIVERSITY

25 of 91

MDP Formulation

25

State transition

Formulation: Markov decision process based safe and scalable trajectory planner

Action set

THE GEORGE WASHINGTON UNIVERSITY

26 of 91

Reachability Analysis

26

Formulation: A reachable set of a system refers to a set that contains all possible future states of the system Data-driven reachability analysis tool known as DryVR [12]

Compute discrepancy parameters

Compute sensitivity parameters

Compute distance between trajectories

[12] Fan, C., Qi, B., Mitra, S., and Viswanathan, M., “DryVR: Data-Driven Verification and Compositional Reasoning for Automotive Systems,” Computer Aided Verification, edited by R. Majumdar and V. Kunčak, Springer International Publishing, Cham, 2017

THE GEORGE WASHINGTON UNIVERSITY

27 of 91

Reachability Analysis

27

THE GEORGE WASHINGTON UNIVERSITY

28 of 91

Working Procedure

28

Compute rewards

Future states

Value

function

Update

states

Best

actions

Goal reached?

Remove aircraft

Aircraft 1

Assign initial and goal states

Update

states

Best

actions

Value

function

Compute rewards

Future states

Goal reached?

Remove aircraft

Aircraft 2

Assign initial and goal states

Update

states

Goal reached?

Best

actions

Value

function

Compute rewards

Future states

Remove aircraft

Aircraft N

Assign initial and goal states

THE GEORGE WASHINGTON UNIVERSITY

29 of 91

Simulation Results

29

Aircraft

Future 10 sec trajectory

Goal state

Traveled path

THE GEORGE WASHINGTON UNIVERSITY

30 of 91

Simulation Result for 8 Agents

30

THE GEORGE WASHINGTON UNIVERSITY

31 of 91

Simulation Result for 16 Agents

31

THE GEORGE WASHINGTON UNIVERSITY

32 of 91

Simulation Result for 32 Agents

32

THE GEORGE WASHINGTON UNIVERSITY

33 of 91

Results

33

Experimental Settings: 25 Experiments for each case Near Mid Air Collision (NMAC) 🡪 152m horizontal and 30m vertical separation [6] All experiments are conducted on Intel(R) Xeon(R) W-2133 cores running at 3.60GHz

[13] S. Chen, A. Evans, M. Brittain and P. Wei, “Integrated Conflict Management for UAM with Strategic Demand Capacity Balancing and Learning-based Tactical Deconfliction”, IEEE Transactions on Intelligent Transportation Systems, vol. 25, no. 8, Jan. 2024.

THE GEORGE WASHINGTON UNIVERSITY

34 of 91

Action Shielding

34

Objective: Filter out actions that result in unsafe states [14]

Future States

Shield

Dead-lock Control Actions

[14] K.nighofer, B., Lorber, F., Jansen, N., and Bloem, R., “Shield synthesis for reinforcement learning,” Leveraging Applications of

Formal Methods, Verification and Validation: Verification Principles. ISoLA 2020, Rhodes, Greece, Springer, 2020

THE GEORGE WASHINGTON UNIVERSITY

35 of 91

Reward Shaping

35

State

Objective: Modify the reward received by the agent to guide the agent toward desired behaviour [15]

Modified

Reward

Reward Shaping

Function

Difference of

Potentials

[15] Ng, A. Y., Harada, D., and Russell, S., “Policy invariance under reward transformations: Theory and application to reward

shaping,” International Conference on Machine Learning (ICML), Vol. 99, 1999

THE GEORGE WASHINGTON UNIVERSITY

36 of 91

Energy-Efficient Trajectory Planning and Mission Feasibility Assessment Framework

[16] Abenezer Taye, Ellis L. Thompson, Peng Wei, Timothy Bonin, and James C. Jones. “Probabilistic evaluation for flight mission feasibility of a small octocopter in the presence of wind,” AIAA AVIATION 2023 Forum.

[17] Abenezer G. Taye and Peng Wei. “Flight mission feasibility assessment of urban air mobility operations under battery energy constraint,” AIAA SCITECH 2024 Forum.

[18] Abenezer G. Taye and Peng Wei. “Energy-efficient trajectory planning and feasibility assessment framework for drone package delivery,” Journal of Aerospace Information Systems (JAIS), 2025.

37 of 91

Motivation: Drone package delivery

37

Problem: Enhance safety and energy-efficiency of drone package delivery operations

Zipline [1]

DroneUp [2]

Amazon Prime Air [3]

Wing [4]

UPS Flight Forward [5]

NASA [6]

THE GEORGE WASHINGTON UNIVERSITY

38 of 91

Motivation: Drone package delivery

38

As UTM progresses in levels of maturity 🡪 thousands of operations simultaneously Need for automated tools that aid in the planning and decision-making process (UAS Operator)

[19] Unmanned Aircraft System (UAS) Traffic Management (UTM) Concept of Operations (ConOps) Version 2.0. 2020

THE GEORGE WASHINGTON UNIVERSITY

39 of 91

Motivation: Drone package delivery

39

Objective: Develop energy-efficient trajectory planner + feasibility assessment framework

THE GEORGE WASHINGTON UNIVERSITY

40 of 91

Objective

40

Develop a pre-departure flight planning and decision-making framework for UAS operations:

Safety between aircraft trajectories

Use reachability analysis based online safety verification

Energy-efficient trajectories

Accounting for wind and several aircraft related factors

Flight mission feasibility assessment

Battery state prediction uncertainty quantification

Realistic package delivery scenario

Assess the performance of the framework

Flight test validation

Partial validation of the feasibility assessment framework

THE GEORGE WASHINGTON UNIVERSITY

41 of 91

Problem Formulation

41

Safe and energy-efficient trajectories

Our problem has two major challenges

Mission feasibility assessment

Pre-departure flight planning and battery energy related flight mission feasibility assessment

Cleared for flight

Hold

THE GEORGE WASHINGTON UNIVERSITY

42 of 91

Contributions

42

Energy-efficient trajectories

Mission feasibility assessment

Flight test validation

THE GEORGE WASHINGTON UNIVERSITY

43 of 91

Hierarchical Framework

43

Working procedure of the proposed framework

Multi-Agent Trajectory

Planning

Mission Feasibility Assessment

THE GEORGE WASHINGTON UNIVERSITY

44 of 91

Upper Layer

44

Trajectory Planner: Markov decision process based safe and energy-efficient trajectory planner

Reward Function

Energy-Efficiency

Reachability Analysis

Package Delivery Scenario

Guidance Model [9]

[20] Corbetta, M., Banerjee, P., Okolo, W., Gorospe, G., and Luchinsky, D. G., “Real-time uav trajectory prediction for safety monitoring in low-altitude airspace,” AIAA AVIATION 2019 forum, 2019

THE GEORGE WASHINGTON UNIVERSITY

45 of 91

Upper Layer

Energy-Efficiency: A set of energy-efficiency reward functions have been developed

THE GEORGE WASHINGTON UNIVERSITY

46 of 91

Energy Cost Function

Objective: Train an energy cost function

Data collection

Future states

Battery simulation

Energy consumption

THE GEORGE WASHINGTON UNIVERSITY

47 of 91

Trajectory Deviation Reward

Objective: Compute the angle between each future state and the shortest path

THE GEORGE WASHINGTON UNIVERSITY

48 of 91

Wind-Aware Planning

Objective: Compute the angle between velocity at each future state and wind field.

THE GEORGE WASHINGTON UNIVERSITY

49 of 91

Results

Package Delivery Scenario: Collected CFD simulated wind data for a region near the city of Boston [10]

[21] Baskar, D., and Gorodetsky, A., “A Simulated Wind-field Dataset for Testing Energy Efficient Path-Planning Algorithms for

UAVs in Urban Environment,” AIAA AVIATION 2020 FORUM, 2020

THE GEORGE WASHINGTON UNIVERSITY

50 of 91

Results

Package Delivery Scenario: Two aircraft conducting a package delivery mission from a depot

THE GEORGE WASHINGTON UNIVERSITY

51 of 91

Results

Power Profile Comparison: Two aircraft conducting a package delivery mission from a depot

THE GEORGE WASHINGTON UNIVERSITY

52 of 91

Lower Layer

52

[22] Daigle, M., and Kulkarni, C. S., “Electrochemistry-based battery modeling for prognostics,” Annual Conference of the PHM Society, Vol. 5, 2013

Objective: Performs battery feasibility assessments for the generated trajectories

THE GEORGE WASHINGTON UNIVERSITY

53 of 91

Lower Layer

53

Objective: Performs battery feasibility assessments for the generated trajectories

Electro-chemical-based model adopted from [12]

A detailed octocopter dynamics adopted from [11]

[23] Ahmed, I., Quinones-Grueiro, M., and Biswas, G., “A high-fidelity simulation test-bed for fault-tolerant octo-rotor control using reinforcement learning,” 2022 IEEE/AIAA 41st Digital Avionics Systems Conference (DASC), IEEE, 2022

[24] Daigle, M., and Kulkarni, C. S., “Electrochemistry-based battery modeling for prognostics,” Annual Conference of the PHM Society, Vol. 5, 2013

THE GEORGE WASHINGTON UNIVERSITY

54 of 91

Final Decisions

Flight Operator

Framework Workflow

Aircraft Type

Flight Planning and Feasibility Assessment

Framework

Battery SoC

Cleared for Flight

Hold the Aircraft

Keep Charging the Aircraft

Wind Forecast

Delivery

Locations

THE GEORGE WASHINGTON UNIVERSITY

55 of 91

Decision Making

55

Objective: Performs battery feasibility assessments and make decisions based of the assessment

Cleared for flight

Hold

THE GEORGE WASHINGTON UNIVERSITY

56 of 91

Decision Making

56

Objective: Performs battery feasibility assessments and make decisions based of the assessment

Cleared for flight

Hold

THE GEORGE WASHINGTON UNIVERSITY

57 of 91

Validation

57

Objective: Hardware validation of the feasibility assessment framework

THE GEORGE WASHINGTON UNIVERSITY

58 of 91

Validation

58

Objective: Hardware validation of the feasibility assessment framework

THE GEORGE WASHINGTON UNIVERSITY

59 of 91

Real-Time Flight Mission Feasibility Assessment and Contingency Management Framework

[*] The content of this chapter has been submitted to the 36th International Conference on Principles of Diagnosis and Resilient Systems (DX 2025) and is currently under review.

60 of 91

Problem Setting

Mission Feasibility

Depot Location

Delivery Location

Delivery Drone

Energy Consumption Model

Future

Trajectory

Learning based

Battery State Predictor

Battery state at

End of Flight

How do we accurately predict battery states under changing conditions, in real time?

How do we build models that are fast enough to meet the feasibility assessment time budget?

Research Problems

THE GEORGE WASHINGTON UNIVERSITY

61 of 91

Related Works

Energy Consumption Model

Future

Trajectory

Learning based

Battery State Predictor

Battery state at

End of Flight

THE GEORGE WASHINGTON UNIVERSITY

62 of 91

Proposed Framework

62

Convert Power Profile

To Current Profile

Learning-based

Battery Model

Battery State Prediction

Initial and Final

Locations

Flight Plan

Generator

Identify Flight

Segments

Power

Consumption

Model

Power Consumption Prediction

Perform Feasibility

Assessment

Proceed?

Reroute

Feasibility Assessment and Decision Making

THE GEORGE WASHINGTON UNIVERSITY

63 of 91

63

Initial and Final

Locations

Flight Plan

Generator

Identify Flight

Segments

Power

Consumption

Model

Power Consumption Prediction

Power and Current Profile Prediction

Power Consumption Model

[25] Gina Sierra, M Orchard, Kai Goebel, and C Kulkarni. Battery health management for small-size rotary-wing electric unmanned aerial vehicles: An efficient approach for constrained computing platforms. Reliability Engineering & System Safety, 2019.

THE GEORGE WASHINGTON UNIVERSITY

64 of 91

64

Initial and Final

Locations

Flight Plan

Generator

Identify Flight

Segments

Power

Consumption

Model

Power Consumption Prediction

Power and Current Profile Prediction

Power Consumption Model

Power-to-Current Conversion

THE GEORGE WASHINGTON UNIVERSITY

65 of 91

Proposed Framework

65

Convert Power Profile

To Current Profile

Learning-based

Battery Model

Battery State Prediction

Initial and Final

Locations

Flight Plan

Generator

Identify Flight

Segments

Power

Consumption

Model

Power Consumption Prediction

THE GEORGE WASHINGTON UNIVERSITY

66 of 91

Proposed Framework

66

Convert Power Profile

To Current Profile

Learning-based

Battery Model

Battery State Prediction

Batteries

Continuous time systems

Dynamics －differential equations

Neural Ordinary Differential Equations

Physics Informed Neural Networks

THE GEORGE WASHINGTON UNIVERSITY

67 of 91

Proposed Framework

67

Convert Power Profile

To Current Profile

Learning-based

Battery Model

Battery State Prediction

Batteries

Continuous time systems

Dynamics －differential equations

Neural Ordinary Differential Equations

Physics Informed Neural Networks

Dataset Construction

Model Training & Evaluation

THE GEORGE WASHINGTON UNIVERSITY

68 of 91

Dataset Generation

68

Full Flight and Mid-Flight Current Profiles

Simulated Voltage Trajectories

[26] Daigle, M., and Kulkarni, C. S., “Electrochemistry-based battery modeling for prognostics,” Annual Conference of the PHM Society, Vol. 5, 2013

THE GEORGE WASHINGTON UNIVERSITY

69 of 91

Neural Differential Equations (Neural ODEs)

69

General Idea Continuous time model that learns the fundamental differential equations of a system Can be thought as a generative time series model

THE GEORGE WASHINGTON UNIVERSITY

70 of 91

Physics Informed Neural Network (PINN)

70

General Idea Incorporating known physics information into the training process Our PINN model is composed of LSTM + Equivalent Circuit Model

Equivalent Circuit Model

THE GEORGE WASHINGTON UNIVERSITY

71 of 91

Results

71

Package Delivery Scenario: Developed a drone package scenario in the Dallas Fort Worth area

THE GEORGE WASHINGTON UNIVERSITY

72 of 91

Results

72

THE GEORGE WASHINGTON UNIVERSITY

73 of 91

Results

73

Package Delivery Scenario: Developed a drone package scenario in the Dallas Fort Worth area

Infeasible

Short

EM3

EM1

THE GEORGE WASHINGTON UNIVERSITY

74 of 91

Results

74

Cleared for flight

Re-route

THE GEORGE WASHINGTON UNIVERSITY

75 of 91

Results

75

Cleared for flight

Re-route

THE GEORGE WASHINGTON UNIVERSITY

76 of 91

Results

76

Package Delivery Scenario: Spatial visualization of voltage predictions along each flight trajectory

THE GEORGE WASHINGTON UNIVERSITY

77 of 91

Results

77

THE GEORGE WASHINGTON UNIVERSITY

78 of 91

Summary

78

Develop trajectory planning and mission feasibility assessment strategies to

enable safe, scalable, energy-efficient, and resilient operations in AAM.

Overarching Research Objective

Energy-efficient trajectory planning and feasibility assessment framework

Real-time feasibility assessment and contingency management framework

Safe and scalable UAM trajectory planner

Chapter 2

Chapter 3

Chapter 4

THE GEORGE WASHINGTON UNIVERSITY

79 of 91

Future Works

79

Energy-efficient trajectory planning and feasibility assessment framework

Real-time feasibility assessment and contingency management framework

Safe and scalable UAM trajectory planner

Chapter 2

Chapter 3

Chapter 4

Incorporating state uncertainty

Cooperative multi-agent setting

Non-homogeneous multi-agent setting

Collision avoidance with static obstacle

Incorporating battery state of health

Model optimization for onboard

deployment

THE GEORGE WASHINGTON UNIVERSITY

80 of 91

Publications

80

Taye, A.G., A. Coursey, M. Quinones-Grueiro, C. Hu, G. Biswas, P. Wei, “Safe to Fly? Real-Time Flight Mission Feasibility Assessment for Drone Package Delivery Operations” submitted to the 36th International Conference on Principles of Diagnosis and Resilient Systems
Taye, A.G., Wei, P., “Energy-Efficient Trajectory Planning and Mission Feasibility Assessment Framework for Drone Package Delivery Operations”, AIAA Journal of Aerospace Information Systems. Mar. 2025.
Taye, A.G., Valenti, R., Rajhans, A., Mavrommati, A., Mosterman, P.J. and Wei, P., “Safe and Scalable Real-Time Trajectory Planning Framework for Urban Air Mobility.” AIAA Journal of Aerospace Information Systems, Aug. 2024.
Taye, A.G. and Wei, P., “Flight Mission Feasibility Assessment of Urban Air Mobility Operations under Battery Energy Constraint”, AIAA SCITECH, Orlando, FL, Jan. 2024.
Taye, A.G., Thompson, E.L., Wei, P., Bonin, T. and Jones, J.C., “Probabilistic Evaluation for Flight Mission Feasibility of a Small Octocopter in the Presence of Wind”, AIAA AVIATION, San Diego, CA, Jun. 2023.
Taye, A.G., Bertram, J., Fan, C. and Wei, P., “Reachability based online safety verification for high-density urban air mobility trajectory planning”, AIAA AVIATION, Chicago, IL, Jun. 2022.

Taye, A.G., Chen, S., Wei, P., “Energy-Aware Strategic Traffic Management for Urban Air Mobility”, Accepted by AIAA SCITECH, Orlando, FL, Jan 2025.
A. Tabrizian, P. Gupta, A. Taye, J. Jones, E. Thompson, S. Chen, T. Bonin, D. Eberle and P. Wei, “Using Large Language Models to Automate Flight Planning under Wind Hazards”, AIAA/IEEE Digital Avionics Systems Conference (DASC), San Diego, CA, Sept. 2024.
Taye, A.G., Wei, P., Pradeep, P., Jones, J.C., Bonin, T., and Eberle, D., “Energy Demand Analysis for eVTOL Charging Stations in Urban Air Mobility”, AIAA AVIATION, Las Vegas, NV, July 2024.
P. Razzaghi, A. Tabrizian, W. Guo, S. Chen, A. Taye, E. Thompson, A. Bregeon, A. Baheri and P. Wei, “A Survey on Reinforcement Learning in Aviation Applications”, Engineering Applications of Artificial Intelligence, vol. 136, part A, Oct. 2024.
Thompson, E.L., Taye, A.G., Guo, W., Wei, P., Quinones, M., Ahmed, I., Biswas, G., Carr, S., Topcu, U. and Jones, J.C., “A survey of eVTOL aircraft and AAM operation hazards.” AIAA AVIATION, Chicago, IL, Jun. 2022.
J. Xu, C. Vu, A. Taye, A. Tabrizian, P. Wei, “Small UAS Landing Site Detection with ArUco Markers and Deep Learning based Computer Vision”, Accepted by AIAA AVIATION, Las Vegas, NV, July 2025.
D. Ding, C. Vu, A. Taye, A. Tabrizian, P. Wei, Z. Zhao, “Synthetic Data Generation for Computer Vision based Autonomous Landing for Small UAS Package Delivery”, Accepted by AIAA AVIATION, Las Vegas, NV, July 2025.

Dissertation Related

Non-Dissertation

Open-Source Code

Chapter 2

Chapter 3

Chapter 4

THE GEORGE WASHINGTON UNIVERSITY

81 of 91

Thank You!�Questions and Comments

Abenezer Taye

abenezertaye@gwu.edu

THE GEORGE WASHINGTON UNIVERSITY

82 of 91

Other Experiences

Community Service

Served as a reviewer:

Journal of Aerospace Information Systems
IEEE Intelligent Transportation Systems
The Aeronautical Journal
AIAA AVIATION and SCITECH

Summer Internships

MathWorks Advanced Research and Technology Office

May 2022 – August 2022

Sensor Fusion and Tracking Toolbox

May 2023 – August 2023

Teaching

Mentored 4 undergrad students in battery-feasibility related research
Teaching assistant for the course MAE 4182: Electro-Mechanical Control Systems (Fall 2024)

THE GEORGE WASHINGTON UNIVERSITY

83 of 91

Results

83

Package Delivery Scenario: Developed a drone package scenario in the Dallas Fort Worth area

Infeasible

Short

EM3

EM1

THE GEORGE WASHINGTON UNIVERSITY

84 of 91

Physics Informed Neural Network (PINN)

84

General Idea Incorporating known physics information into the training process Our PINN model is composed of LSTM + Equivalent Circuit Model

Equivalent Circuit Model

THE GEORGE WASHINGTON UNIVERSITY

85 of 91

Physics Informed Neural Network (PINN)

85

General Idea Incorporating known physics information into the training process Our PINN model is composed of LSTM + Equivalent Circuit Model

THE GEORGE WASHINGTON UNIVERSITY

86 of 91

Physics Informed Neural Network (PINN)

86

General Idea Incorporating known physics information into the training process Our PINN model is composed of LSTM + Equivalent Circuit Model

THE GEORGE WASHINGTON UNIVERSITY

87 of 91

Physics Informed Neural Network (PINN)

87

General Idea Incorporating known physics information into the training process Our PINN model is composed of LSTM + Equivalent Circuit Model

THE GEORGE WASHINGTON UNIVERSITY

88 of 91

Physics Informed Neural Network (PINN)

88

General Idea Incorporating known physics information into the training process Our PINN model is composed of LSTM + Equivalent Circuit Model

THE GEORGE WASHINGTON UNIVERSITY

89 of 91

Neural Differential Equations (Neural ODEs)

89

General Idea Continuous time model that learns the fundamental differential equations of a system Can be thought as a generative time series model

State

Encoder

Decoder

THE GEORGE WASHINGTON UNIVERSITY

90 of 91

Neural Differential Equations (Neural ODEs)

90

General Idea Continuous time model that learns the fundamental differential equations of a system Can be thought as a generative time series model

① Deterministic

Latent space = Voltage space

② Voltage prediction

THE GEORGE WASHINGTON UNIVERSITY

91 of 91

Neural Differential Equations (Neural ODEs)

91

General Idea Continuous time model that learns the fundamental differential equations of a system Can be thought as a generative time series model

THE GEORGE WASHINGTON UNIVERSITY