A maneuverability study on a wheeled bin management robot in tree fruit orchard environments
MetadataShow full item record
An important activity during harvesting is bin management, which traditionally requires heavy use of seasonal skilled labor. Due to increased labor cost and uncertainty with labor availability in recent years, tree fruit industry in the Washington State can benefit from robotic solutions to combat with these problems. A “bin-dog” system is a robotic bin management system used in orchard. The goal of this research was to develop an automated bin-dog system of good maneuverability for effective fruit bin management in typical PNW tree fruit orchard environment. To fulfil this goal, a conceptual prototype of a bin-dog system was developed and tested. The proposed bin-dog system adopted an innovative “go-over-the-bin” concept, which allowed the conceptual system to manage bins in a five-step process. To obtain effective maneuverability on bin-dog, a coordinate control system for a four-wheel-independent-steering (4WIS) system was developed to independently control the rotational speeds and steering angles of all its wheels. Field tests in a commercial orchard showed that the bin-dog system could follow the process to replace a full bin on an aisle with an empty bin. To provide guideline information for navigating bin-dog with the 4WIS system, the influence of major factors on path tracking performance using four proposed steering modes were investigated through a series of field tests. The results indicated that longitudinal driving speed, look-ahead distance and/or steering mode could be carefully selected for different paths and tasks to achieve a high path tracking accuracy and small spatial requirement. A steering strategy selection algorithm was developed to generate and determine an optimized steering strategy to complete designated operations including steering back to center line of an aisle, steering into an aisle from headland and bin loading on an aisle in orchard environment. The field test indicated that the algorithm was able to generate applicable steering strategies, correctly determine whether a steering strategy would lead to collision with boundaries of worksite, and select an optimized steering strategy which was collision-free and had shortest theoretical total travel distance.