Data-Driven Motion Planning for Robots

Mon, 01 Jan 2024 00:00:00 +0000

We systematically survey how data-driven approaches — deep neural networks, reinforcement learning, and large language models — can be applied to robot motion planning, and explore new directions in planning methods. By combining the strengths of classical search-based planning and learning-based planning, we aim at motion planning that achieves both adaptability and generalization in the real world.

Multi-Robot Coordination with Large Language Models

Mon, 01 Jan 2024 00:00:00 +0000

We integrate the commonsense reasoning of large language models (LLMs) with structured optimization such as linear programming and dependency graphs, enabling cooperative task planning by multiple robots. The LLM decomposes natural-language tasks, while inter-robot dependencies and constraints are solved through optimization to produce feasible cooperative behaviors even under real-world constraints.

Action-Planning | Takato Horii | Osaka University

Data-Driven Motion Planning for Robots

Multi-Robot Coordination with Large Language Models