# A sensor-to-pattern calibration framework for multi-modal industrial collaborative cells

**Authors:** Daniela Rato, Miguel Oliveira, Vítor Santos, Manuel Gomes, Angel Sappa  
**arXiv:** [2210.10365](https://arxiv.org/abs/2210.10365)  
**Date:** 2022-10-19  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2210.10365v1/

## Key points

- Collaborative robotic industrial cells involve robots working alongside human operators
- Safety is crucial in these environments and requires a comprehensive understanding of the workspace
- Multiple sensors are needed to capture information from different viewpoints within the cell
- Integrating data from these sensors is challenging due to the complex nature of the system
- The calibration process becomes difficult because of the large number of sensors and limited overlapping fields of view
- A sensor-to-pattern calibration framework is proposed to address this issue
- The methodology can handle various types of sensors, including RGB and Depth cameras, as well as LiDARs
- Precise calibration is achieved for a collaborative cell with multiple cameras and LiDARs using this approach
- Pattern-based calibration techniques effectively align sensor data for an accurate perception of the workspace
- This research enhances safety by enabling reliable fusion of information from multiple sensors in collaborative robotic cells.

## Summary

Collaborative robotic industrial cells are workspaces where robots collaborate with human operators. Safety is of utmost importance in these environments, and to ensure it, a comprehensive understanding of the space where the collaborative robot operates is necessary. This requires equipping collaborative cells with a wide range of sensors that capture information from multiple modalities, covering the entire work volume. However, integrating data from all these sensors poses a challenge due to the complex nature of the system. The calibration process becomes particularly difficult because of the large number of sensors and modalities involved, as well as the limited overlapping fields of view between them. These sensors are strategically positioned to capture different viewpoints within the cell. To address this issue, this paper presents a sensor-to-pattern calibration framework that enables accurate calibration of complex systems such as collaborative cells in a single optimization procedure. The proposed methodology can handle various types of sensors including RGB and Depth cameras, as well as LiDARs. The results demonstrate that this methodology achieves precise calibration for a collaborative cell comprising three RGB cameras, a depth camera, and three 3D LiDARs. By leveraging pattern-based calibration techniques, this approach effectively aligns the sensor data to provide an accurate perception of the workspace. This research contributes to enhancing safety in collaborative robotic industrial cells by enabling reliable fusion of information from multiple sensors. With improved extrinsic calibration, these systems can better understand their surroundings and facilitate seamless collaboration between robots and human operators.