Localisation method of strawberry nursery greenhouse robots based on visual infrared beacons

Achieving accurate localisation of mobile robots in greenhouse environments is challenging due to interfering factors, such as iron seedbeds, steel frame structures, and fluctuating ambient light conditions. This study proposes a localisation method using monocular vision and infrared LED beacons, which employ specific wavelengths to reduce ambient light impact. In addition, a combination of threshold segmentation and region-growing segmentation is used to extract the infrared LED beacons rapidly from the background. The monocular vision ranging principle is adopted to determine the relative distance between the camera mounted on a mobile platform and the beacons. Also, a trilateral localisation mechanism is used to calculate the mobile platform’s position. Experimental tests on a tracked mobile robot verified the proposed method. In the static localisation experiments, the proposed method has an average absolute error of 5.5 cm and a maximum absolute error of 8.1 cm. In dynamic localisation experiments, at robot speeds of 0.2 m/s and 0.4 m/s, the maximum drift errors are 5.7 cm and 6.0 cm, with average absolute drift errors of 2.0 cm and 2.2 cm, respectively. The results confirm the effectiveness of the proposed method in the precise localisation of robots in challenging greenhouse environments.