Authors: Arvi Jonnarth, Jie Zhao, Michael Felsberg

Published on: June 29, 2023

Impact Score: 8.15

Arxiv code: Arxiv:2306.16978

Summary

• What is new: A novel reinforcement learning approach with an egocentric map representation and total variation-based reward term for coverage path planning.
• Why this is important: The challenge of planning a path that covers a confined area’s entire space, especially when the environment is unknown and changing.
• What the research proposes: Using reinforcement learning with newly developed components like egocentric map representation and a unique reward function to efficiently learn coverage paths.
• Results: The approach outperforms existing RL-based methods and specialized techniques across various CPP scenarios.

Technical Details

Technological frameworks used: Reinforcement Learning

Models used: Egocentric map representation, Total variation-based reward

Data used: Simulated unknown environments

Potential Impact

Robotics for lawn mowing, cleaning, search-and-rescue operations, and surveillance.

Want to implement this idea in a business?

We have generated a startup concept here: FrontierMaps.