Bioinspired online learning onboard a drone
Abstract
Real-time learning is essential for agents to adapt to their specific tasks and environments, making the underlying algorithm for real-time learning crucial for achieving true autonomy--agent operation without any human intervention. Discovering a better algorithm which is closer to what is observed in nature would have significant applications for both neuroscience and robotics-AI. In neuroscience, it would address the fundamental question of brain structure formed by the core task of adaptation to anticipated demands. By knowing what circuitry ought to do, one can better direct a search for how the corresponding algorithm is implemented. For robotics and AI, the development of an efficient real-time learning algorithm could lead to better autonomous agents capable of learning and adapting to their specific tasks and environments, significantly boosting efficiency in civil and military applications.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 05, 2025
- Source ID
- FA86552417061
Entities
People
- Guido de Croon
Organizations
- Air Force Office of Scientific Research
- Delft University of Technology
- United States Air Force