# Legged Robots for Object Manipulation: A Review

**Authors:** Yifeng Gong, Ge Sun, Aditya Nair, Aditya Bidwai, Raghuram CS, John Grezmak, Guillaume Sartoretti, Kathryn A. Daltorio  
**arXiv:** [2303.16865](https://arxiv.org/abs/2303.16865)  
**Date:** 2023-03-29  
**Summary page:** https://www.summarizepaper.com/en/paper/2303.16865v1/

## Key points

- Legged robots have potential in manipulating objects in challenging environments
- Existing research on legged robotics has focused on traversing difficult terrains, but some demonstrations show they can also move objects
- The review categorizes existing works into four main manipulation methods: object interactions without grasping, manipulation with walking legs, dedicated non-locomotive arms, and legged teams
- Each method has different design features and autonomy levels
- Quantitative comparisons are provided for the range of possible relative sizes of manipulated objects with respect to the robot based on a few simplifying assumptions
- New directions for research in legged robot manipulation include multifunctional limbs, terrain modeling, or learning-based control
- Advancements could support new deployments in challenging indoor/outdoor scenarios such as warehouses/construction sites, preserved natural areas, and especially home robotics.
- Continued research is needed to develop more advanced legged robotic systems capable of performing complex tasks in human-centric environments.

## Summary

This review article titled "Legged Robots for Object Manipulation: A Review" explores the potential of legged robots in manipulating objects in challenging environments that are inaccessible to humans. While most research on legged robotics has focused on traversing difficult terrains, some demonstrations have shown that these robots can also move objects as a way of doing tangible work. The authors aim to collect and learn from these examples to organize the work done so far in the community and highlight interesting avenues for future research. The review categorizes existing works into four main manipulation methods: object interactions without grasping, manipulation with walking legs, dedicated non-locomotive arms, and legged teams. Each method has different design features and autonomy levels, which are illustrated by available examples in the literature. The authors further provide quantitative comparisons for the range of possible relative sizes of manipulated objects with respect to the robot based on a few simplifying assumptions. The examples presented suggest new directions for research in legged robot manipulation such as multifunctional limbs, terrain modeling, or learning-based control. These advancements could support new deployments in challenging indoor/outdoor scenarios such as warehouses/construction sites, preserved natural areas, and especially home robotics. Overall, this review provides valuable insights into how legged robots can be designed to manipulate various types of objects collaboratively or independently. It highlights the potential applications of these robots beyond just traversing difficult terrains and emphasizes the need for continued research in this field to develop more advanced legged robotic systems capable of performing complex tasks in human-centric environments.