1. Overview

With aging population, farmers abandoning their farmlands have been increased in Japan. On the other hand, agricultural management is becoming increasingly organized and incorporated with the increase of certified farmers. The consolidation of farmlands by these farmers has increased the area cultivated by each faming organization. In order to sustain efficient and stable farm management, it is necessary to improve productivity and profitability, which requires agricultural machinery to become more efficient and labor-saving. With the promotion of ICT-based smart agriculture, next-generation agriculture, Kubota has developed an Agri Robo tractor that uses GNSS and enables unmanned automatic operation, in order to accomplish higher efficiency and precision and reduce more workers and labors.

2. Derail of Technology

For unmanned operation of a tractor, we have established the following three techniques by mounting a GNSS positioning device, electro-hydraulic power steering, a terminal monitor (operation panel), an inter-vehicle communication device, a laser scanner and an ultrasonic sonar (Fig. 1) and controlling these devices in an integrated manner.

Due to that the behavior of a vehicle successively varies with skids or ground pits or projections in a field, it is necessary to continuously steer the vehicle even during driving straight. The developed unmanned tractor (Agri Robo tractor) has two types of control targets, “running to a target direction (smallest orientation deviation)” and “running to a target route (smallest position deviation),” and a control algorithm that determines a steering output by a combination of these targets. These elements allow the tractor to perform high-precision traveling in any field conditions. During controlled operation along a specific target route, both of position and orientation deviations remain stable at relatively small values, but immediately after the change of an operation sequence, such as shift from turning to straight running, a large position deviation is instantly generated. For this reason, without any control, a steering output becomes excessive, which finally reduces convergence to the target route. Therefore, we have developed a variable gain algorithm in which the vehicle becomes closer to the target route, increases its orientation gain and gradually decreases its approach angle, and allowed the vehicle to continuously perform the best control according to the state of the body. Fig. 2 shows a relationship between a distance from a target line and gain.

When the tractor turns, a positional relation with a start position for the next line, a line orientation or the vehicle can be calculated. Based on this relation, the tractor generates a virtual turning route and turns along the route. While the virtual turning route is generated based on this positional relationship and the turning operation is executed along the route, the tractor may protrude from the field depending on the azimuth of the traveling line to the ridge. In this case, the turning route is recalculated so that the tractor can turn within the minimum possible space (Fig. 6). In addition, retry control to decelerate when deviation from the turning route exceeds a certain level during a turning operation and generate a new turning route further improves turning precision.

We have established a technique for avoidance of collision with an obstruction as part of a safety system. A laser scanner is mounted on the right and left and rear sides of the vehicle, so that the tractor stops before collision with the obstruction. The selected laser scanner can scan one plane at wide angles, so is suitable for detection of any remote obstruction. Due to that a blind spot is generated under the scanning plate of the laser scanner within easy reach of the tractor, however, an ultrasonic sonar is deployed to allow for detection of a low obstruction. Fig. 4 shows a conceptual image of ranges of obstruction detection.

Fig. 1  Unmanned operation equipment

Fig. 2  Distance from target line and gain

Fig. 3  Turning route

Fig. 4  Conceptual image of ranges of obstruction detection

Tomofumi Fukunaga*1

Yushi Matsuzaki*2

Hiroki Suga*3

Kotaro Yamaguchi *4

Akisato　Hori*4

^*1       Member, Kubota North America Corporation (USA)

^*2       Kubota Corporation (64 Ishizukitamachi, Sakai-ku, Sakai, Osaka 590-0823 Japan)

^*3       Full member, Kubota Corporation (64 Ishizukitamachi, Sakai-ku, Sakai, Osaka 590-0823 Japan)

^*4       Kubota Corporation (1-1-1 Hama, Amagasaki, Hyogo 661-8567 Japan)