Autonomous Localization and Navigation for Agricultural Robots in Greenhouse

Agricultural robots are an effective way to solve the increasingly prominent shortage of agricultural labor. To meet the needs of practical applications, agricultural robots must be able to locate autonomously and move from one place to another automatically. A map-based autonomous localization and navigation system for low-speed agricultural robot in medium-sized greenhouse is developed in this paper. By the perception of on-board sensors, the developed system can build and update environmental map timely, estimate robot’s pose precisely, locate and navigate autonomously. To effectively solve the SLAM problem, RBPF-based nonparametric solution is adopted, and its evolution is also described systematically and concisely. Relevant tests were carried out in greenhouse and laboratory corridor respectively to verify the feasibility of the developed system, and the results show that the autonomous localization and navigation system can not only build consistent environmental map, but also autonomously plan the moving paths to goal positions. By taking calibrated landmarks as benchmarks, the performance of autonomous localization is evaluated, and the corresponding evaluation tests show that the autonomous localization precision in both environments can reach centimeter level. The centimeter level localization precision of the developed system can fully meet the requirements of the operations in the greenhouse. Furthermore, tests shows that the unfavorable environment, such as uneven, unstructured and illumination uncertain, resulted in lower autonomous localization precision in the greenhouse than in the laboratory corridor.