A Review of Scene Representations for Robot Manipulators

TL;DR

This paper reviews various scene representations used by robot manipulators, emphasizing how different sensing modalities and tasks influence the internal world models that enable manipulation and interaction.

Contribution

It provides a comprehensive overview of scene representations derived from real-world sensing specifically for robot arms, highlighting their diversity and application in downstream tasks.

Findings

Scene representations vary widely based on robot type and sensing modality.

Most representations are designed to support manipulation and collision avoidance.

The review categorizes and compares different approaches in the field.

Abstract

For a robot to act intelligently, it needs to sense the world around it. Increasingly, robots build an internal representation of the world from sensor readings. This representation can then be used to inform downstream tasks, such as manipulation, collision avoidance, or human interaction. In practice, scene representations vary widely depending on the type of robot, the sensing modality, and the task that the robot is designed to do. This review provides an overview of the scene representations used for robot manipulators (robot arms). We focus primarily on representations which are built from real world sensing and are used to inform some downstream robotics task.