Title: Accuracy assessment of a GPS-based auto-guidance system in an agricultural vehicle using computational vision methods
Authors: Rigoberto Castro Castro; Ricardo Yassushi Inamasu; Maira Martins Da Silva
Addresses: Faculty of Engineering, Mechatronics Engineering Department, Universidad Tecnológica Centroamericana (UNITEC), Tegucigalpa 11101, Honduras ' Embrapa Instrumentation São Carlos, R. 15 de Novembro 1452, São Carlos-SP, 13560-970, Brazil ' São Carlos School of Engineering, University of São Paulo (USP), Av. Trab. Sancarlense 400, São Carlos, 13566-590, Brazil
Abstract: A real improvement in efficiency and productivity has been successfully achieved in precision agriculture using auto-guidance systems. The global positioning system (GPS) real-time kinematic (RTK) system, which allows for centimetre accuracy, is one alternative for implementing such systems. However, geographic positioning errors, vehicle dynamics, agricultural devices, and field environment conditions may influence the performance of GPS-based autonomous agricultural vehicles. Measuring the vehicle position using cameras, lasers, odometers, and ultrasonic sensors can aid this influence assessment. This work aims to propose a methodology to assess the accuracy of auto-guidance systems under actual field conditions using computer vision methods. The pinhole camera method was used to map vehicle location by processing a checkerboard image in the field. The proposal is validated by performing several field tests. The use of computer vision methods can be an accurate alternative to evaluate auto-guidance systems if devices, procedures, and parameters are appropriately selected and calibrated.
Keywords: computational vision; image processing; RTK GPS navigation; precision agriculture; autonomous driving system; vehicle dynamics; agricultural vehicles; vehicle location.
DOI: 10.1504/IJHVS.2022.123244
International Journal of Heavy Vehicle Systems, 2022 Vol.29 No.1, pp.95 - 106
Received: 10 Jul 2021
Accepted: 24 Nov 2021
Published online: 06 Jun 2022 *