This is the project webpage of ORLA*: Mobile Manipulator-Based Object Rearrangement with Lazy A*. Source Codes and Paper Will be Available Soon.

Problem Introduction

In this paper, we investigate the problem of Mobile Robot Tabletop Rearrangement (MoTaR), where a robot with a mobile base is tasked to move objects from an initial arrangement to a desired goal arrangement on a confined tabletop.

We study two different scenarios of MoTaR:

	EE (End-Effector) Scenario	MB (Mobile Base) Scenario
Examples
Features	Rearrangement on a small table, where the robot can stay at a fixed position.	Rearrangement on a large table, where the robot needs to travel around for pick and places.
Considered Costs	End-effector (EE) travel distance + # pick-n-places	Mobile base (MB) travel distance + # pick-n-places

Highlights and Contributions

[Studied Problem] We seek time-optimal solutions to Mobile Robot Tabletop Rearrangement (MoTaR).

[A* with Lazy Buffer Allocation] We propose Object Rearrangement with Lazy A* (ORLA*), which employs the idea of lazy buffer allocation into the A* framework to search for optimal rearrangement plans minimizing the cost function.

[Pose Stability Estimation] To estimate the feasibility of a buffer pose, especially when we want to temporarily place an object on top of others, we propose a learning model, StabilNet.