On The Comparison of Model-Based and Model-Free Controllers in Guidance, Navigation and Control of Agricultural Vehicles

TY - CHAP

T1 - On The Comparison of Model-Based and Model-Free Controllers in Guidance, Navigation and Control of Agricultural Vehicles

AU - Kayacan, Erkan

AU - Kayacan, Erdal

AU - Chen, I-Meng

AU - Ramon, Herman

AU - Saeys, Wouter

PY - 2018/2/8

Y1 - 2018/2/8

N2 - In a typical agricultural field operation, an agricultural vehicle must be accurately navigated to achieve an optimal result by covering with minimal overlap during tillage, fertilizing and spraying. To this end, a small scale tractor-trailer system is equipped by using off the shelf sensors and actuators to design a fully autonomous agricultural vehicle. To alleviate the task of the operator and allow him to concentrate on the quality of work performed, various systems were developed for driver assistance and semi-autonomous control. Real-time experiments show that a controller, which gives a satisfactory trajectory tracking performance for a straight line, gives a large steady-state error for a curved line trajectory. On the other hand, if the controller is aggressively tuned to decrease the tracking error for the curved lines, the controller gives oscillatory response for the straight lines. Although existing autonomous agricultural vehicles use conventional controllers, learning control algorithms are required to handle different trajectory types, environmental uncertainties, such as variable crop and soil conditions. Therefore, adaptability is a must rather than a choice in agricultural operations. In terms of complex mechatronics systems, e.g. an agricultural tractor-trailer system, the performance of model-based and model-free control, i.e. nonlinear model predictive control and type-2 neuro-fuzzy control, is compared and contrasted, and eventually some design guidelines are also suggested.

AB - In a typical agricultural field operation, an agricultural vehicle must be accurately navigated to achieve an optimal result by covering with minimal overlap during tillage, fertilizing and spraying. To this end, a small scale tractor-trailer system is equipped by using off the shelf sensors and actuators to design a fully autonomous agricultural vehicle. To alleviate the task of the operator and allow him to concentrate on the quality of work performed, various systems were developed for driver assistance and semi-autonomous control. Real-time experiments show that a controller, which gives a satisfactory trajectory tracking performance for a straight line, gives a large steady-state error for a curved line trajectory. On the other hand, if the controller is aggressively tuned to decrease the tracking error for the curved lines, the controller gives oscillatory response for the straight lines. Although existing autonomous agricultural vehicles use conventional controllers, learning control algorithms are required to handle different trajectory types, environmental uncertainties, such as variable crop and soil conditions. Therefore, adaptability is a must rather than a choice in agricultural operations. In terms of complex mechatronics systems, e.g. an agricultural tractor-trailer system, the performance of model-based and model-free control, i.e. nonlinear model predictive control and type-2 neuro-fuzzy control, is compared and contrasted, and eventually some design guidelines are also suggested.

UR - https://www.ebay.de/itm/Type-2-Fuzzy-Logic-and-Systems-Studies-in-Fuzziness-and-Soft-Computing-/183150553872?hash=item2aa49fab10

U2 - 10.1007/978-3-319-72892-6_3

DO - 10.1007/978-3-319-72892-6_3

M3 - Book chapter

SN - 9783319728919

VL - 362

T3 - Studies in Fuzziness and Soft Computing

SP - 49

EP - 73

BT - Studies in Fuzziness and Soft Computing

A2 - John, R.

A2 - Hagras, H.

A2 - Castillo, O.

PB - Springer

ER -