Skydio
Autonomy Intern, Computer Vision Team
Perception for autonomous obstacle avoidance on the X10 drone: semantic segmentation and monocular and stereo depth estimation for thin objects.
Guillem Casadesús Vila
An aerospace engineer from Barcelona interested in robotic space exploration, autonomous and distributed space systems, optimization, and machine learning
About
I am a PhD candidate in Aeronautics and Astronautics at Stanford University, working on lunar navigation and autonomy in the Navigation and Autonomous Vehicles (NAV) Lab, advised by Prof. Grace Gao. My work covers lunar Positioning, Navigation, and Timing (PNT), rover perception, mapping and autonomy, simulation, and humanoid locomotion. The navigation, mapping and planning stack developed in our lab won first place in NASA's Lunar Autonomy Challenge.
In 2022, I completed my Bachelor's thesis on dynamic resource management in satellite communications as a visiting student in the System Architecture Group, Department of Aeronautics and Astronautics, Massachusetts Institute of Technology (MIT). In parallel, our MIT team developed a mission concept for producing propellants on Mars from in-situ resources, which won first place at NASA's RASC-AL competition.
I obtained a BSc in Aerospace Engineering and a BSc in Telecommunications Engineering at Universitat Politècnica de Catalunya (UPC) under the program offered by the Interdisciplinary Higher Education Center (CFIS). My studies were funded by a “la Caixa” Foundation Fellowship and a Cellex Foundation–CFIS UPC Fellowship. During high school, I started the Gadget Robotics team.
Soccer, skiing, scuba diving, and traveling are my favorite activities to take a break from work.
Research and Projects
Lunar Communications and Navigation Satellite Constellations
Navigation and Autonomous Vehicles – Stanford
With dozens of lunar missions planned in the coming decade, there is growing interest in deploying satellite constellations that provide communications and Positioning, Navigation, and Timing (PNT) services around the Moon. We study how to design and operate them: receiving Earth-GPS signals in lunar orbit, sensor fusion for orbit determination and time synchronization, scheduling of PNT services, delay-tolerant networking, and a lunar surface station that supports orbit determination. LuPNT, our open-source simulator for lunar communications and PNT, supports these studies. Our work on orbital sensor fusion for autonomous orbit determination and time synchronization won Best Paper in Space Communications and Navigation at the IEEE Aerospace Conference 2025.
Lunar Surface Navigation and Autonomy
Navigation and Autonomous Vehicles – Stanford
Rovers and humanoids will work alongside astronauts on the Moon, and autonomy, hazard avoidance and navigation are critical capabilities for them, especially while lunar navigation constellations are not yet deployed. We develop the layers of an autonomy stack: feature matching for visual SLAM, real-time semantic 3D Gaussian splatting maps with depth estimation, SLAM anchored to orbital elevation maps, and hazard detection and path planning. We test them in photorealistic lunar simulation, whose renders we compare against Apollo, Chang'e-3 and Artemis images, and on lunar testbed, Earth analog and real lunar data. We also study humanoid locomotion under lunar gravity, terrain and soil. The navigation, mapping and planning stack developed in our lab won first place in NASA's Lunar Autonomy Challenge.
[NASA news] [Lunar Autonomy Challenge paper] [Humanoid preprint] [Publications]
Deep Reinforcement Learning for Trajectory Design in Cis-Lunar Space
Stanford Computer Science
Interest in operating spacecraft between the Earth and the Moon, in cis-lunar space, is growing. Computing low-thrust trajectories there in real time is hard, because traditional optimization methods are computationally expensive. In this class project, we trained deep reinforcement learning policies that output low-thrust trajectories and their control inputs in real time.
Dynamic Resource Management for Communication Satellites
System Architecture Group – MIT
Modern high-throughput satellites with digital payloads and phased arrays have thousands of configurable variables, too many to allocate by hand. We developed optimization methods, including integer linear programming, that assign frequencies to satellite beams and re-plan them as user demand changes.
Martian In-Situ Resource Utilization from Subsurface Water Ice
Engineering Systems Laboratory – MIT
NASA plans crewed missions to the Moon and Mars, and producing propellants and consumables from local resources is a key enabler for them. Our concept for NASA's RASC-AL competition uses trucks working in tandem to extract subsurface water ice and produce methalox at scale, with a focus on reliability and scalability. It won first place overall.
[Publications] [MIT News] [MIT AA News] [NASA story]
Distributed Satellite Systems Simulator and Autonomous Operations
Nanosatellite and Payload Laboratory – UPC
Earth observation missions increasingly rely on distributed satellite systems, which need dedicated simulation tools for their design and operation. We developed a distributed satellite systems simulator to validate new mission architectures and to study satellite networking and autonomous operations, including learning-based scheduling.
Machine Learning and Distributed Computing Models
Workflows and Distributed Computing Group – BSC
Machine learning libraries are difficult to run on high-performance computing clusters. I developed machine learning algorithms for dislib, the Distributed Computing Library, a Python library built on PyCOMPSs, a programming model and runtime for clusters and clouds.
Autonomous Robotics Competitions
Gadget Robotics Team
I started the Gadget Robotics team in high school. We designed and built autonomous robots for line-following races, mini-sumo and humanoid fights, and competed in the Spanish National Robotics League.
Experience
Stanford University
Research Assistant, Navigation and Autonomous Vehicles Lab
Autonomy stack for lunar rovers combining visual SLAM, 3D Gaussian splatting mapping and semantic perception, and LuPNT, an open-source C++ simulator for lunar orbital and surface navigation.
Stanford University
Teaching Assistant and Lecturer, AA 272: Global Positioning Systems
Taught the graduate course for three years.
NASA Jet Propulsion Laboratory
Research Intern, Spacecraft Autonomy and Simulation
Navigation algorithms for multi-satellite telescopes and the High-Energy X-ray Probe.
Massachusetts Institute of Technology
Research Intern, System Architecture Group
Uncertainty-aware spectrum optimization for high-throughput satellite communications.
Barcelona Supercomputing Center
Research Intern, Workflows and Distributed Computing Group
Machine learning algorithms for high-performance and distributed computing.
Nano-Satellite and Payload Laboratory, UPC
Research Intern, Distributed Satellite Systems
Satellite simulator and learning-based scheduling for large constellations.
Heat and Mass Transfer Technological Center, UPC
Research Intern, Turbulent Flow in Aerodynamics
Direct numerical simulations of turbulent flow.
Service
- Member, Vice Provost for Graduate Education (VPGE) Faculty Advisory Committee, Stanford
- Member, Aeronautics and Astronautics Department Student Advisory Council, Stanford
- President and Treasurer, Stanford Catalan Association
- President, Stanford AIAA student branch
Publications
* equal contribution
- G. Casadesus Vila and G. Gao, “Modular Autonomy Stack for Lunar Surface Rover Missions,” IEEE Aerospace Conference, Big Sky, MT, March 2027. (Abstract accepted)
- G. Casadesus Vila and G. Gao, “Comprehensive Evaluation of Front-End Feature Matching for Lunar Visual SLAM,” IEEE Aerospace Conference, Big Sky, MT, March 2027. (Abstract accepted) [arXiv]
- G. Casadesus Vila, S. A. Deevi and G. Gao, “Do Earth-Trained Humanoid Controllers Walk on the Moon? Gravity, Terrain, and Soil Transfer,” arXiv preprint. [arXiv]
- A. Dai, G. Casadesus Vila and G. Gao, “Visual SLAM with DEM Anchoring for Long-Range Lunar Surface Navigation,” IEEE Aerospace Conference, Big Sky, MT, March 2026. [paper] [arXiv]
- G. Casadesus Vila, A. Dai and G. Gao, “Semantic Segmentation and Depth Estimation for Real-Time Lunar Surface Mapping Using 3D Gaussian Splatting,” IEEE Aerospace Conference, Big Sky, MT, March 2026. [paper]
- A. Dai, A. Wu, K. Iiyama, G. Casadesus Vila, K. Coimbra, T. Deng and G. Gao, “Full Stack Navigation, Mapping, and Planning for the Lunar Autonomy Challenge,” NAVIGATION: Journal of the Institute of Navigation, 2026 (Submitted). [arXiv]
- G. Casadesus Vila and G. Gao, “Sensor Fusion for Autonomous Orbit Determination and Time Synchronization in Lunar Orbit,” IEEE Aerospace Conference, Big Sky, MT, 2025. Best Paper in Space Communications and Navigation. [paper]
- A. Dai, A. Wu, K. Iiyama, G. Casadesus Vila, K. Coimbra, T. Deng and G. Gao, “Full Stack Navigation, Mapping, and Planning for the Lunar Autonomy Challenge,” Proceedings of the Institute of Navigation GNSS+ Conference (ION GNSS+ 2025), Baltimore, MD, September 2025. [paper]
- G. Casadesus Vila and G. Gao, “Lunar Surface Station to Support Lunar Positioning, Navigation, and Timing Services,” Proceedings of the Institute of Navigation GNSS+ Conference (ION GNSS+ 2025), Baltimore, MD, September 2025. [paper]
- G. Casadesus Vila, K. Iiyama and G. Gao, “LuPNT: An Open-Source Simulator for Lunar Communications, Positioning, Navigation, and Timing,” IEEE Aerospace Conference, Big Sky, MT, 2025. [paper]
- G. Casadesus Vila and G. Gao, “Markov Decision Processes for Scheduling Lunar PNT Services,” Proceedings of the Institute of Navigation GNSS+ Conference (ION GNSS+ 2024), Baltimore, MD, 2024. [paper]
- K. Iiyama*, G. Casadesus Vila*, and G. Gao, “Contact Plan Optimization and Distributed State Estimation for Delay Tolerant Satellite Networks,” IEEE Aerospace Conference, Big Sky, MT, 2024. [paper]
- C. Agia, G. Casadesus Vila, S. Bandyopadhyay, et al., “Modeling Considerations for Developing Deep Space Autonomous Spacecraft and Simulators,” IEEE Aerospace Conference, Big Sky, MT, 2024. [paper] [arXiv]
- J. J. G. Luis, S. Aliaga, G. Casadesus Vila, N. Pachler, E. F. Crawley, and B. G. Cameron, “Frequency Plan Design for Multibeam Satellite Constellations Using Integer Linear Programming,” IEEE Transactions on Wireless Communications, 2024. [paper] [PDF]
- K. Iiyama*, G. Casadesus Vila*, and G. Gao, “LuPNT: Open-Source Simulator for Lunar Positioning, Navigation, and Timing,” Proceedings of the Institute of Navigation GNSS+ Conference (ION GNSS+ 2023), Denver, CO, 2023. [paper]
- P. Morrissey, K. K. Madsen, R. Amini, D. Dawson, B. Grefenstette, A. Abramovici, S. Bandyopadhyay, G. Casadesus Vila, et al., “The High-Energy X-ray Probe (HEX-P),” SPIE UV, X-Ray, and Gamma-Ray Space Instrumentation for Astronomy XXIII, San Diego, CA, 2023. [paper]
- G. Casadesus Vila, J. A. Ruiz-de-Azua and E. Alarcon, “Toward Autonomous Cooperation in Heterogeneous Nanosatellite Constellations using Dynamic Graph Neural Networks,” International Astronautical Congress, Baku, Azerbaijan, 2023. [paper] [arXiv]
- G. Casadesus Vila, J. J. G. Luis, N. Pachler, E. F. Crawley, and B. G. Cameron, “Dynamic Frequency Assignment for Mobile Users in Multibeam Satellite Constellations,” International Astronautical Congress, Paris, France, 2022. [paper] [PDF] [arXiv]
- C. Gentgen, G. Casadesus Vila, G. C. Lordos, J. Posada, M. Long, L. Nagareddy, J. Kambhampaty, M. T. Have, M. Hoying, J. A. Hoffman, and O. L. D. Weck, “BART & MARGE: A Resilient and Scalable Architecture for In-Situ Propellant Production on Mars Using Large Trucks,” AIAA ASCEND 2022, Las Vegas, NV, 2022. [paper]
Education
Stanford University
MSc and PhD in Aeronautics and Astronautics
Navigation and Autonomous Vehicles (NAV) Lab. Advisor: Prof. Grace Gao.
Massachusetts Institute of Technology
BSc Thesis in Aeronautics and Astronautics
Thesis: Dynamic Frequency Assignment for Mobile Users in Multibeam Satellite Constellations. Advisor: Prof. Edward F. Crawley.
Universitat Politècnica de Catalunya – CFIS
BSc in Aerospace Engineering (rank 1/60) and BSc in Telecommunications Engineering (rank 1/163)
Speaking and Media
Presenter at the [Bay Area Robotics Symposium]
Spotlight talk and poster on lunar surface navigation and autonomy
Presenter at the [Stanford Center for AeroSpace Autonomy Research (CAESAR) Annual Meeting]
Presenter at the [Stanford Center for AI Safety Annual Meeting]
Best Paper Award at the IEEE Aerospace Conference [NAV Lab News]
Best Paper in Space Communications and Navigation for “Sensor Fusion for Autonomous Orbit Determination and Time Synchronization in Lunar Orbit”
Featured in [NASA News]
First place overall at the NASA Lunar Autonomy Challenge
Stanford University [Spotlight]
Profile covering robotics projects, lunar PNT research in the NAV Lab and Mars propellant production
Presenter at AIAA ASCEND
Paper on the Mars propellant production concept
Presenter at the International Astronautical Congress
Paper on frequency assignment for satellite constellations
Interview in [la Veu de la Segarra]
Personal projects, “la Caixa” Fellowship award, and grad school
Featured in [NASA News], [MIT News] and [MIT AeroAstro News]
First place at the NASA RASC-AL competition
Presenter at the NASA RASC-AL Final Forum
Mars propellant production architecture
Presenter at [Malakabot Robotics]
Design and construction of a humanoid robot
Creator of [Gadget Robotics Team]
Team that competed in the Spanish National Robotics League
Featured in [UPF News]
First place at the XI UPF Research and Engineering Awards
Featured in [Ara] & [laPedrera Foundation]
First place at the Exporecerca Jove research fair
More about me
