# Navigate-and-Seek: a Robotics Framework for People Localization in Agricultural Environments

**Authors:** Riccardo Polvara, Francesco Del Duchetto, Gerhard Neumann, Marc Hanheide  
**arXiv:** [2107.03850](https://arxiv.org/abs/2107.03850)  
**Date:** 2021-07-08  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2107.03850v1/

## Key points

- The paper addresses the potential for productivity gains in agriculture through robotic automation
- Accurate detection and localization of humans is crucial for collaboration between robots and human workers
- The proposed framework uses a topological particle filter (TPF) for accurate individual localization and tracking of workers in farm environments
- The framework integrates information from heterogeneous sensors and combines local active sensing with global localization using affordable IoT GNSS devices
- Improved accuracy in picker localization compared to prior work is demonstrated through multi-sensor observations on the topological level and active perception strategies
- Validation of the approach is conducted in topologies designed for deploying robotics fleets in real farm environments
- Effective integration of sensor data and active perception strategies significantly enhance worker localization accuracy
- The framework has the potential to improve productivity and efficiency in farming operations involving human-robot collaboration.

## Summary

The paper titled "Navigate-and-Seek: a Robotics Framework for People Localization in Agricultural Environments" addresses the potential for productivity gains in the agricultural domain through the introduction of robotic automation. The authors highlight the importance of detecting and localizing humans accurately in order to facilitate collaboration between mobile robots and human workers. To address this challenge, they propose an approach based on a topological particle filter (TPF) that enables accurate individual localization and tracking of workers in a farm environment. The proposed framework integrates information from heterogeneous sensors and combines local active sensing, which leverages a robot's onboard sensing using a Next-Best-Sense planning approach, with global localization using affordable IoT GNSS devices. By combining multi-sensor observations on the topological level with active perception through the Next-Best-Sense approach, the authors demonstrate improved accuracy in picker localization compared to prior work. The validation of the proposed approach is conducted in topologies specifically designed for deploying robotics fleets to support fruit pickers in real farm environments. The results show that by effectively integrating sensor data and employing active perception strategies, it is possible to enhance worker localization accuracy significantly. Overall, this paper presents a comprehensive robotics framework that addresses the crucial task of people localization in agricultural environments. By leveraging topological particle filtering and combining local active sensing with global localization techniques, this framework has significant potential for improving productivity and efficiency in farming operations involving human-robot collaboration.