BECAS BGI2026 - Advanced Learning, Artficial Intelligence and Control Laboratory. Advanced Robotics Internship: UAV Autonomy & Safe Navigation - ESTADOS UNIDOS DE AMERICA (Houston)
- Código Oferta:14772
- Proyecto: UNIVERSITY OF HOUSTON / BECAS BGI2026 - Advanced Learning, Artficial Intelligence and Control Laboratory. Advanced Robotics Internship: UAV Autonomy & Safe Navigation - ESTADOS UNIDOS DE AMERICA (Houston) / ESTADOS UNIDOS DE AMERICA / Houston
- País: ESTADOS UNIDOS DE AMERICA
- Municipio: Houston
- Entidad: UNIVERSITY OF HOUSTON
- Tipo de contrato: Indefinido
- Duración: Mínimo 6 meses
- Salario mensual: 1.785 euros brutos
- Observaciones: Our long-term research goal is to develop coordinated safe and collision-free motion planning for teams of Unmanned Aerial Systems (UAS) operating in dynamic, GNSS-denied, and poorly understood complex environments. To achieve this goal, in this project, we will focus on two main research objectives: * (RO1) development of a centralized, high-level task planning module for one-tomany interaction and closed-loop task execution; * (RO2) development of a lowlevel receding-horizon sampling-based and collision-free kinodynamic motion planner. The task of UAS path planning for teams of small vehicles involves determining an optimal coordinated route from an initial location to a target point while avoiding collisions with obstacles and maintaining operational safety at all times. Assigning a dedicated pilot to each vehicle does not scale with swarm size, it is inefficient and prone to human error. On the other hand, managing swarm behavior without any human supervision at all may not be preferred, because it lacks situational awareness and complete knowledge of the entire operation, prerogatives of mission commanders, as well as it exposes the swarm to safety risks. in the proposed research objective RO1, we will develop a module for centralized one-to-many interactions exploiting LLMs for task planning and swarm coordination. In this framework, users will issue high-level commands in natural language and a previously trained LLM model will translate them into structured machine executable API calls. We will also enable the LLM to be triggered whenever an event requiring a new motion plan occurs, for instance, when a fault in any of the drones is detected or when new vehicles join the team. We will also enable telemetry data to be sent back to the LLM in a closed-loop fashion, allowing the model to decide what tasks should be prioritized and if replanning is necessary. To the best of our knowledge, this is the first investigation of using a language model to interact in closed-loop with both humans and a team of UAVs. Once the structured execution plans are generated to fulfill the desired tasks, the second research objective RO2 in this proposal is concerned with exploiting the execution plans to create low-level path plans for poorly understood complex environments. State-of-the-art motion planning for navigation in unknown environments is typically done as an online pipeline that tightly couples 3D mapping, local planning, trajectory optimization, and tracking. By exploiting optimal sampling-based motion planning, optimal control, and safety assurance filters we will develop a low-level receding-horizon samplingbased safe and collision-free kinodynamic motion planner. Our proposed planner will effectively detect any potential unsafe control inputs during the trajectory planning process, and it will provide minor adjustments when needed to guarantee safe and successful mission completion. In our lab, we emulate satellite dynamics on small aerial platforms, both in simulation and experimentally with actual vehicles, to verify and test the developed close proximity satellite autonomy control algorithms. Candidates will be responsible for conducting research and writing scientific papers for dissemination of results. Candidates will work under the direct supervision and will be affiliated with the Advanced Learning, Artificial Intelligence and Control Lab.
- Nivel Académico: Licenciatura o Grado (C5-8)
- Fecha Alta: 2026-06-25
- Fecha Inicio: 2026-07-23
- Titulación: Ingeniería Informática, Ingeniería de Sistemas de Control, Ingeniería Eléctrica, Ingeniería Electrónica, Ingeniería Robótica, Ingeniería Aeroespacial, Ingeniería Mecánica, Machine Learning
- Observaciones Idioma:
- Informática: Phyton
© 2026 Fomento de San Sebastián.
