Design and field testing of a polygonal paddy infield path planner for unmanned tillage operations

Abstract

Path commands, obtained through a traveling route strategy for auto-steering control in a test field, are essential for the operation of an autonomous tractor from a given starting point to an end destination. Planning an agricultural field operation can increase the work rate and quality of field tilling, particularly in complex fields, minimizing the time loss and non-working traveling costs. With the diversification of farming products and environments, it is critical to seek feasible paths that have a good balance between time efficiency and path length, with the consideration for irregularities of field boundaries. However, the existing field path planning algorithms do not guarantee global optimality because traditional planners generate paths in rectangular fields and not in all the possible geometrical route generation for infield and headland turning found in polygonal fields. To overcome this drawback, this paper presents a novel and practical infield path planner that can generate infield coverage routes and headland turnings for polygonal paddy fields. An infield route-planning program was developed to automatically generate an X-type turning, pattern-based coverage path map that provides guidance lines for autonomous tillage tractors executing efficient operations in polygonal fields. The planning parameters related to the equipment characteristics of the tractor and implement variables, waypoint-setting variables, and field boundaries entered into the developed infield route-planning program can provide useful information to users in terms of selecting the most appropriate solution based on all related field operational capacities such as field traveled distance, field time, and field efficiency. A small-scale prototype of an autonomous tillage tractor equipped with the proposed infield path planner and a slip estimation-based path-tracking algorithm was built and tested in Korean paddy fields with two different shapes of farm-field boundaries. The field tests performed using the autonomous tractor in a trapezoidal field and an L-shaped field showed tracking errors of 14.9 and 21.8 cm, respectively; field efficiencies of 62.1–70.6%; and ratios of effective operation areas of 91.4–92.7%. These measurement data were similar to those obtained from a human-operated tillage method, implying that the infield path planner developed here and the slip estimation-based path-tracking controller can lead a tractor with acceptable accuracy and performance.