Integrated Navigation Models of a Mobile Fodder-Pushing Robot Based on a Standardized Cow Husbandry Environment
HighlighIntegrated navigation models for a two-wheel robot were specifically developed for a semi-enclosed environment.A combination of Kalman filter and fuzzy control system was developed with mathematical models.Real-time pose estimate and adjustment of perturbances due to feeding cows and fodder resistance were achieved.Abstract. As part of welfare feeding, standardized feeding is commonly used for cows in confined operations. Due to the strict facility requirements, smart mobile robots have been specifically developed to address these semi-enclosed environments. Their navigation is based on electromagnetic sensors with magnetic tapes, which does not easily allow route changes and other abilities afforded by the newer integrated sensors and Global Navigation Satellite System (GNSS) guidance packages available on large agricultural machinery in outdoor environments. This article proposes a system of integrated navigation using multiple sensors, which was used for a two-wheel-drive robot operating in the standardized environment of a cow husbandry facility. The developed system combined incremental encoders, ultrasonic sensors, and a gyroscope to determine parameters such as course angle and covered distance. A fuzzy self-adaption Kalman filter was applied to integrate these parameters and estimate the robot pose, so that the robot could achieve real-time course adjustment during operation. Experimental trials indicated that the real-world route was highly consistent with the set route. Moreover, the cross-track error was =0.10 m at a travel velocity of 0.2 m s-1, indicating that perturbances due to feeding cows and fodder resistance had little interference on the movement of the robot, and the models were robust and accurate. This novel integrated sensing system with a fuzzy self-adaption Kalman filter and derived models was able to guide real-time robot operations in a modern cow husbandry environment without the need for magnetic tapes. Keywords: Kalman filter, Integrated navigation, Motion models, Pose estimate, Welfare feeding.