This repository presents an undergraduate research project accepted under the TÜBİTAK 2209-A University Students Research Projects Support Program.
The project investigates whether a spacecraft operating in Low Earth Orbit can autonomously assess collision risk, plan an avoidance maneuver and subsequently generate a new mission orbit to maintain operational continuity.
Unlike approaches focused solely on immediate collision avoidance, the proposed research also considers post-maneuver mission replanning and decision-making under delayed or unavailable ground-station communication.
The increasing number of active satellites and debris objects in Low Earth Orbit creates growing operational risks for spacecraft.
Conventional collision-avoidance operations generally depend on ground-based tracking, human analysis and commands transmitted from a ground station. Communication delays, limited contact windows and the increasing scale of satellite constellations can restrict the effectiveness of this process.
This project investigates a simulation-based autonomous decision-support architecture intended to reduce this dependency while preserving mission safety and continuity.
- Model representative spacecraft and debris encounters in Low Earth Orbit.
- Evaluate collision risk under different orbital scenarios.
- Generate appropriate collision-avoidance maneuver alternatives.
- Replan the mission orbit following an avoidance maneuver.
- Consider fuel consumption, mission deviation and decision time.
- Provide maneuver alternatives to the ground station.
- Investigate autonomous decision-making when ground communication is unavailable.
- Compare autonomous and ground-assisted operational scenarios.
flowchart TD
A[Orbital Scenario] --> B[Collision-Risk Assessment]
B --> C[Avoidance-Maneuver Planning]
C --> D[Mission-Orbit Replanning]
D --> E[Ground-Assisted or Autonomous Decision]
E --> F[Performance Evaluation]
The research is based on a simulation environment representing spacecraft and debris motion in Low Earth Orbit.
The planned study includes:
- Orbital propagation using TLE-compatible methods
- Scenario-based spacecraft and debris modelling
- Guidance-based avoidance-maneuver generation
- Reinforcement-learning-assisted decision making
- Post-maneuver mission-orbit replanning
- Ground-assisted and autonomous decision scenarios
- Statistical analysis of simulation outputs
Candidate artificial-intelligence approaches include reinforcement-learning algorithms such as PPO and DQN. Final algorithm selection will be determined through feasibility studies and comparative testing.
Specific algorithm combinations, reward-function formulations and implementation details are not publicly disclosed while the research remains active.
The proposed system will be evaluated using metrics including:
- Collision-avoidance success
- Mission-continuity rate
- Fuel and maneuver efficiency
- Deviation from the planned mission orbit
- Decision time
- Number of required maneuvers
- Performance under communication-delay scenarios
- Robustness under different debris-density conditions
- Yigit Alp Gulsen — Project applicant and student project lead; orbital simulation, system integration and project documentation
- Hacer Durmus — Reinforcement-learning agent development
- Sude Nur Yildiz — Software testing and validation
As the project applicant and student project lead, my responsibilities include:
- Contributing to the research methodology and project planning
- Developing the orbital simulation environment
- Supporting autonomous-planning architecture development
- Coordinating system-level integration and final evaluation
- Analysing simulation outputs
- Preparing technical documentation and project presentations
| Item | Status |
|---|---|
| TÜBİTAK 2209-A Application | Accepted |
| Research and Literature Review | In Progress |
| Simulation Environment | In Progress |
| Autonomous Agent Development | Planned / In Progress |
| Scenario-Based Validation | Planned |
| Academic Publication | Planned |
The status table will be updated as validated research outputs become available.
The project is expected to produce:
- A simulation-based autonomous collision-avoidance demonstrator
- A post-maneuver mission-replanning framework
- Comparative results for different operational scenarios
- An undergraduate research report
- A national peer-reviewed publication submission
- An international conference-paper submission
These outputs are research objectives and should not be interpreted as completed publications or finalized results.
- Program: TÜBİTAK 2209-A University Students Research Projects Support Program
- Application Period: 2025 First Call
- Institution: Necmettin Erbakan University
- Department: Aerospace Engineering
- Research Area: Autonomous Spacecraft Operations and Space-Traffic Safety
- Project Lead: Yigit Alp Gulsen
The simulation code, training environment, reward-function design and research data are not publicly available while the project remains active.
Validated and publication-safe technical materials may be released following completion of the research and academic-publication process.
For further information about the project scope, research objectives or my technical contributions, please feel free to contact me.
- Email: yigitalpgulsen26@gmail.com
- LinkedIn: Yigit Alp Gulsen
- LinkedIn: Hacer Durmuş
- LinkedIn: Sude Nur Yıldız
This repository serves as a public technical overview of the research project. The original TÜBİTAK application document, unpublished research materials and implementation-specific details are intentionally not included.