# Design and Evaluation of Motion Planners for Quadrotors

**Authors:** Yifei Simon Shao, Yuwei Wu, Laura Jarin-Lipschitz, Pratik Chaudhari, Vijay Kumar  
**arXiv:** [2309.13720](https://arxiv.org/abs/2309.13720)  
**Date:** 2023-09-24  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2309.13720v1/

## Key points

- Significant advancements in quadrotor motion planning over the past decade
- Two-stage approach involving front-end and back-end stages
- Front-end determines optimal path considering geometric and kinematic/input constraints
- Back-end refines path using objective function, robot dynamics, and limitations
- Lack of systematic methods for designing front and back ends across environments
- Modular approach proposed for designing software planning stack
- Parameterized set of environments for systematic evaluation of two-stage planners
- Performance of planning algorithms demonstrated across various environments through simulations and experiments
- Planning/evaluation stack and parameterized environments open-sourced at https://github.com/KumarRobotics/kr_mp_design

## Summary

Significant advancements have been made in the field of quadrotor motion planning over the past decade. These successful approaches typically involve two stages: a front-end that determines the optimal path by considering geometric constraints and kinematic or input constraints to define the trajectory's homotopy class; and a back-end that refines the path using an appropriate objective function, taking into account the robot's dynamics and state/input limitations. However, there is a lack of systematic methods or guidelines for designing both front and back ends across various environments. Additionally, existing literature does not thoroughly assess how trajectory planning algorithms perform under different levels of environmental complexity. To address these gaps, a modular approach to designing the software planning stack has been proposed in this study. The researchers offer a parameterized set of environments that allow for systematic evaluation of two-stage planners. These parametrized environments enable researchers to test different front and back-end planners based on environmental clutter and complexity levels. Through simulations and experimental results, the performance of selected planning algorithms has been demonstrated across a range of environments. Furthermore, the planning/evaluation stack and parameterized environments have been open-sourced to facilitate more comprehensive studies in quadrotor motion planning. Interested individuals can access these resources at https://github.com/KumarRobotics/kr_mp_design. This collaborative effort by Yifei Simon Shao, Yuwei Wu, Laura Jarin-Lipschitz, Pratik Chaudhari, and Vijay Kumar showcases a novel approach to designing and evaluating motion planners for quadrotors with potential implications for future research in this field.