María Gabriela Mendoza

I am a Ph.D. candidate in Mechanical Engineering at UC Berkeley, where I am fortunate to be advised by Prof. Shankar Sastry and mentored by Prof. Chinmay Maheshwari and Dr. Victoria Tuck (postdoctoral researcher, UPenn).

My research focuses on the design and control of multi-agent systems operating in highly uncertain and dynamic environments, with applications in Advanced Air Mobility, humanitarian robotics, human behavior modeling, and cyberphysical systems.

Before Berkeley, I spent five years as a Research Scientist at Lockheed Martin, contributing to projects ranging from hypersonic aerothermal design early in my career to AI and robotics research in later years, often in collaboration with academic partners. I am also deeply committed to mentoring and inclusive engineering education, and previously served as engineering faculty at Cañada College.

I hold a B.S. in Mechanical Engineering from the University of Southern California and an M.S. in Aeronautics and Astronautics from Purdue University.

I design algorithms for modeling, analyzing, and controlling complex interactions in societal-scale multi-agent systems. I develop frameworks for coordination and decision-making in cyber-physical systems where physical dynamics, communication constraints, uncertainty, and strategic interactions are tightly coupled.

I emphasize modeling and simulation that reflect real-world operating conditions: time-varying, unknown environments, uncertainty, low observability and limited communication among heterogeneous agents, and system-level constraints, rather than idealized assumptions.

My interests lie in humanitarian and safety-critical applications such as disaster response, transportation, resource allocation, and planning for underserved populations, spanning autonomous aerial mobility, rescue robotics, and multi-robot systems.

I combine learning-based approaches with control theory, optimization, game theory, and mechanism design, always selecting the method appropriate to the system and task at hand. A growing thread of my work studies foundation models in multi-agent roles: as generative agents simulating sequential human decision-making, and as high-level coordinators for heterogeneous robot teams. My goal is to design systems that are effective, reliable, and principled, with safety, interpretability, and performance guarantees where possible.

• May 2026 Presented “Decentralized Ergodic Coverage Control in Unknown Time-Varying Environments” at the AAMAS 2026 Autonomous Robots and Multirobot Systems (ARMS) Workshop in Paphos, Cyprus (oral).
• Apr 2026 Presented “Hierarchical Generative Agents for Simulating Sequential Human Behavior” at the ICLR 2026 Workshop on Multi-Agent Learning in the Era of Generative AI in Rio de Janeiro, Brazil (poster).
• Spring 2026 Graduate Student Instructor for EE 290: Learning-Enabled Multi-Agent Systems at UC Berkeley.
• Oct 2025 Presented “Coordinated Autonomous Drones for Human-Centered Fire Evacuation in Partially Observable Urban Environments” at the IEEE Global Humanitarian Technology Conference (GHTC) in Boulder, Colorado.

* indicates equal contribution. For an updated list, see my Google Scholar.

Journal articles

Conference papers

Workshop papers

• May 2026 AAMAS 2026, Paphos, Cyprus
  “Decentralized Ergodic Coverage Control in Unknown Time-Varying Environments”
• Spring 2026 EE 290: Learning-Enabled Multi-Agent Systems, UC Berkeley (guest lecture)
  “RESCUE: Resilient Exploration and Search Coordination of UAVs in Unknown Environments”
• 2025–2026 DARPA ANSR Program Meetings
  “Coordinated Autonomous Drones for Human-Centered Fire Evacuation in Partially Observable Urban Environments”
• Oct 2025 IEEE GHTC 2025, Boulder, Colorado
  “Coordinated Autonomous Drones for Human-Centered Fire Evacuation in Partially Observable Urban Environments”
• Sep 2024 Encuentro de Mujeres Latinas en STEM del Área de la Bahía (invited speaker)
  “Un renacer científico en el extranjero: una travesía entre lo profesional, la readaptación cultural y el proceso de autoconocimiento”

• Spring 2026 EE 290 — Learning-Enabled Multi-Agent Systems
  Graduate Student Instructor, UC Berkeley
• 2022–2023 ENGR 100 — Introduction to Engineering
  Instructor, Cañada College (San Mateo Community College District)
• 2022–2023 ENGR 215 — Computational Methods for Engineers and Scientists
  Instructor, Cañada College (San Mateo Community College District)

• 2022–2023 Engineering Faculty · Cañada College
• 2017–2022 Research Scientist · Lockheed Martin
  AI/ML and robotics research; hypersonic systems; multi-university collaborations

• New Foundation models as coordinators for heterogeneous robot teams
  Studying how LLMs can serve as high-level coordinators for heterogeneous multi-robot systems operating under uncertainty
• RESCUE Resilient Exploration and Search Coordination of UAVs in Unknown Environments
  Multi-UAV coordination to assist humans and emergency responders during disasters
• Theory Decentralized ergodic coverage control in unknown time-varying environments
• Allocation Decentralized dynamic task assignment with temporal constraints and uncertainty
• Behavior Simulating sequential human decision-making during disasters in dynamic environments