Fundamental Fund; FF
Physically Interactive AI and Infrastructure
Basic Research:
Towards AI Systems that Perceive, Reason, and
Act in the Real Word
OVERVIEW
Traditional AI often overlooks the importance of physical embodiment, including the dynamic interplay between the body, brain (AI), and environment, for effective data acquisition and real-world interaction. This project aims to address this limitation by developing bio-inspired physical limbed bodies for AI systems, enabling them to not only efficiently acquire data but also actively engage with and adapt to their surroundings, akin to “Living Machine-like AI”. This advancement will make AI more practical, efficient, and resilient for a wide range of real-world applications.
Understanding the Concept: Why bio-inspired limbed bodies?
The limbed bodies for AI systems offer several key advantages:
Effective Data Acquisition: The bodies can actively engage with their surroundings, enabling them to gather data in a more efficient and targeted manner.
Enhanced Adaptation: The bodies can learn from their interactions with the environment. They can make AI systems more resilient to unexpected challenges and changes in their surroundings.
Improved Interaction with Humans: The bodies can interact with humans in a more intuitive and natural way, facilitating collaboration and communication.
Potential Applications: i) Search and rescue, ii) environmental monitoring, iii) transportation, and iv) personal assistance.
In summary, bio-inspired limbed bodies offer a promising approach for developing AI systems that can effectively interact with the real world, learn from experience, and adapt to changing conditions.The Garden
OBJECTIVES
Obj. 1: Design and develop generic bio-inspired limbs with hybrid rigid-soft structures for AI systems, which can be used to construct different bio-inspired limbed bodies, e.g., a bio-inspired exoskeleton (body with a single limb) for human-machine collaboration and rehabilitation and a bio-inspired legged robot (body with multiple limbs) for industrial inspection, search and rescue, and environmental monitoring.
Obj. 2: Equip these limbed bodies with sensors and actuators for effective interaction with the real-world environment, efficient data acquisition, and motion intelligence.
Obj. 3: Integrate AI algorithms to enable and demonstrate complex actions (locomotion) and adaptation through dynamical body-AI-environment interactions.
Obj. 4: Test and evaluate the performance of the embodied AI systems (AI with bio-inspired physical bodies) in real-world scenarios (e.g., exoskeleton for human walking assistance, legged robot for industrial inspection and complex terrain navigation).