R&D Project
Goal 3 R&D Projects (Selected in 2020)Co-evolution of Human and AI-Robots to Expand Science Frontiers
Project manager (PM)HARADA KanakoProfessor, Graduate Schools of Medicine/Engineering, The University of Tokyo
Summary of the project
This project aims to develop AI-robots conducting scientific experiments in a challenging environment (e.g. in a hazardous atmosphere, or in a micro-scale setup), while interacting with scientists as a peer scientist. AI-robots and scientists will change their mutual interactions and do tries-and-errors together to handle objects or work in environments that they have never experienced with. The proposed concept will realize the discovery of scientific principles and solutions by AI-Robots in the science fields by 2050.
Produced by: University of Tokyo
Vision of AI-robot scientists in 2050
Milestone for 2030
The project will develop AI-robots that estimate and reflect the intentions and thoughts of scientists based on their past activities to autonomously perform scientific experiments that could not be done by human scientists alone.
Milestone for 2025
The project will develop AI-robots that automatically interpret a large set of experimental results and propose hypotheses to scientists that humans would not be able to come up with, enabling scientists to formulate new strategies to perform scientific experiments that could not be done by human scientists alone.
R&D theme progress reports
R&D theme structure of the project
AI-robot scientists that autonomously explore science and work as a peer scientist require the body whose physical capabilities are beyond those of human scientists in terms of accuracy, precision, and dexterity, and the brain whose processing capabilities are beyond those of human scientists in terms of the data size and data modalities to handle, and the body and brain need to be integrated so that the AI-robots scientists can autonomously explore science without being pre-programmed by human scientists. Laboratory automation is to automatically conduct pre-programmed experiments, and thus the time required for scientific discovery can be shortened, but the quality of discovery would be the same. The AI-robot scientists with degrees of autonomy are to enable experiments previously thought impossible and alter the quality of scientific discovery by compensating human scientists. The project involves researchers in many disciplines, including both engineers and scientists. The collaborating scientists use prototypes for scientific applications while giving feedback to engineers.
<Topics>
- (1) Body of the AI-robot Scientists
- Robotic platforms and micro-robotic tools will be developed to enable accurate, precise, and dexterous manipulation beyond the physical capabilities of human scientists.
- (2) Brain of the AI-robot Scientists
- Science AI to interpret experimental results and propose hypotheses, Robot AI to interpret experimental manipulation and propose manipulation strategies, and mathematical foundations to abstract knowledge and skills will be developed.
- (3) Scientific applications
- Scientists will use prototypes of AI-robots to conduct experiments.
Leader's institution
The University of Tokyo
R&D institutions
Nagoya University, The Australian National University, Osaka University, Ritsumeikan University, The University of Tokyo, Tohoku University, Kyusyu University, Institute of Science Tokyo, SyntheticGestalt
PDF Download
- Summary of the project (326MB)
- Progress Report (Download all) (797KB)