Research
SYSTEM / ONE SYSTEM
Designing soft and hard as one system.
In harmony between soft and hard, multiple agents cooperate with one another and move wisely as one system in the real world. In particular, we design flexible systems that extend to adapt to the real world, and explore intelligent systems that solve real-world problems.
- Multi-agent
- Cooperation
- Multi-objective
- Route planning
SOFT / ROBOT INTELLIGENCE
Design the intelligence
FITSoftware that fits the hardware
We build a robot's intelligence around its body. Reinforcement learning designs policies by which many vessels and robots move together while avoiding one another.
- Reinforcement learning
- Explainable AI
- Distributional RL
- LLM
- Distributional Reinforcement LearningComing soon
- Explainable AI for Robot Control
- LLM-Driven Reinforcement LearningComing soon
HARD / ROBOT BODY
Design the body
FITHardware that fits the software
We consider the body in which intelligence works best, so that machines move reliably where the ground cannot be seen.
- Self-localisation
- Point-cloud matching
- Craters
- Planetary explorers
- Field robots
- Machine learning
- Uncertainty