It is increasingly common to see news about autonomous vehicles capable of taking care of all farming operations with tractors, drones for spraying, seeders and harvesters that work without human interference. But it's important to understand where we are and how we got here. There is a lot of technological complexity involved, but a simple and succinct approach is still possible. Every advance is the result of a series of gradual evolution events that bring new opportunities for new advances. Thus, by understanding the concepts involved in the steps of this process, we can understand the entire course of technology evolution.
At the beginning of the 1990s, with the release of the use of GPS (North American global positioning system), the first targeting systems via GNSS (global navigation satellite systems) appeared in the USA. With the use of light bars in agricultural and forestry aviation, “flags” were replaced, which were field employees who signaled the line to be followed by the pilot. Obviously it was an unhealthy, risky activity with limited accuracy. Such was the impact on agricultural aviation that, within a few years, light bars were already being widely used in machines and agricultural operations on the ground, especially those with greater difficulty in delimiting route lines, as is the case with the application of solids by broadcast. and spraying.
The light bars, as the name suggests, initially consisted of a panel with a row of LEDs that signaled to the operator whether the machine would be in the correct position on the operating line, turning the LEDs on to the left or right and indicating to the operator the misalignment. The interface has evolved into a screen, which shows the operator the route range, especially more effective when operating on curves. The path is drawn based on the first pass, generating infinite parallel passes of a width defined by the operator.
From the evolution of light bars, the first tractors with automatic steering emerged, which keeps the tractor aligned with the paths established for parallel passes, based on the same technology as light bars. It is still necessary to avoid obstacles and headland maneuvers, but the operator is free to keep the tractor aligned during passes, reducing fatigue and increasing the quality of parallel operations.
For this, positioning signals are necessary, which in the vast majority of cases are given exclusively by GNSS, which are processed to command the steering of the wheelset and thus keep the tractor aligned on the established route. Thus, automatic steering systems for tractors and other agricultural vehicles consist of a GNSS receiver, a data processing unit (computer), an inertial sensor to compensate for the tractor's inclination and signal sudden changes in direction, an angle sensor steering in the steering wheel and an actuator in the steering system.
The differences in the quality of this steering system basically come from the GNSS receivers and steering actuators. But one must also consider the existence or not of some of the other components, especially inertial and steering angle sensors.
The action can be directly on the steering or on the steering wheel. Electro-hydraulic valves and, more recently, electric actuators coupled to the steering system, are more accurate and efficient compared to steering wheel actuators or directly on the steering column. These are subject to set clearances and require greater response time for direction corrections. However, they are cheaper and can be attached to vehicles that were not designed with an automatic steering system, and are recommended for operations that do not require high accuracy, such as hauled input applications, for example.
The GNSS receiver specifications define the accuracy that can be achieved by the system. These specifications define the range of value to be paid, which is certainly higher the greater the desired accuracy. This increase in accuracy must be accompanied by the steering actuator, since steering wheel actuators do not represent gains proportional to the improvement in positioning provided by more accurate GNSS receivers.
An important aspect for automatic steering systems, especially in tractors, is the alignment between it and the machine or implement to be pulled or mounted. On curved paths or with a lateral inclination, misalignment occurs, especially in drag equipment, interfering with the desired path. There are solutions on the market that minimize this problem, some with actuators in the equipment, called active systems, and others without actuators, or passive systems.
In passive systems, the alignment of the towed equipment is prioritized on the route, calculating the tractor's route to compensate for its deviations, without the need for actuators to govern the orientation of the equipment that performs the operation. However, this causes the tractor and the equipment to establish different paths and cannot be used in cases of traffic control. The equipment follows the routes, but the tractor does not, which causes traffic in unwanted locations and damage to crops in crop management operations.
Active systems consist of actuators to correct equipment deviations and allow it and the tractor to follow the same path. Depending on the type of machine and coupling, this actuator can be hydraulic (piston), performing lateral movement of the hitch bar in relation to the tractor or even transverse movement in conjunction with the three-point hitch. For soil preparation equipment, one or more smooth anchoring discs are used, which serve as a reference for alignment and hydraulic actuators like the previous ones. In the case of equipment with wheels (trailers, seeders, planters, fertilizer machines, sprayers), hydraulic actuators can be used to steer directly on the wheels. Most of these solutions, both passive and active systems, require two GNSS antennas, one for the equipment and one for the tractor.
Until the mid-2000s, equipment was able to follow predefined routes and maintain alignment, but was still dependent on the operator. At the same time, work was already being done on the prospect of having them autonomous. To do this, it is necessary to have, among others, the ability to execute maneuvers autonomously, which is not trivial.
Some important items to automate this step are the cabin controls for the exit and re-entry of the set into the field. This automation took place in the 1990s, especially in Europe, where crops are predominantly smaller and maneuvers are more demanding. These are the commands to lift and turn off the drive of the machine coupled to the tractor, with speed, rotation and other adjustments, and immediately after the maneuver, repeat this process in reverse order to re-enter the field carrying out the operation.
The other process is exactly the maneuver route, which can be of different formats, but all require space and time to execute. But as the routes are already defined, the return does not necessarily need to happen in the next pass, allowing optimizations in this sense.
Likewise, more efficient route planning has already been widely adopted in some types of crops. This planning is done based on topographic data of the crops and optimizing the use of spaces and machines. By storing these routes, the vehicle operates even more independently of the operator.
Continuously and with a certain speed, progress is being made towards effectively autonomous vehicles. Much of the technology applied here is based on the automation of road vehicles. In these, decision-making programming is basically divided into three processing layers: controlling layer, responsible for the basic commands, braking, accelerating, steering; sequencing layer, responsible for creating the sequence of actions to be carried out, for example the sequence of exit steps, headland maneuver and re-entry into the field; and deliberative layer, in which the predefined route is read and possible route deviations due to obstacles are calculated, always checking whether the main route is being followed.
This type of processing is based on artificial intelligence, which allows the input of data from several different sources and seeks the most efficient output possible. Thus, it is capable of improving the vehicle's decision-making the more it is used, through machine learning processes.
At the same time, electronic injection systems in engines and automatic transmission brought the possibility of controlling the speed and power of vehicles. Added to this is data from multiple sensors spread across the machines to inform the system and increase the efficiency of operations. It is possible, for example, to vary the speed of cereal harvesters based on the volume of biomass or productivity of the crop so that it always works at the best feeding rate for maximum efficiency and with limited grain loss.
Everything indicates that the market is moving towards full automation in mechanized agricultural systems in the coming decades. To achieve this, technologies still need to evolve so that the operator is no longer necessary, he will become a manager. However, gradually the entire production system will have actions carried out by machines, without the need for human supervision.
Environments with controlled conditions, such as protected crops, for example, will be the first to present viable commercial solutions for fully autonomous vehicles. The paths to be followed are defined by the spatial structure of the crops, which facilitates control of the equipment and its actions in space. Furthermore, the availability of accurate environmental data can also aid decision-making by robots.
The integration between machines, of a mechanized system, in the field, is already pursued, and when fully implemented, it will make the actions of one machine in operation directly influence the others. When harvesting grains or sugarcane, for example, the flow of transshipments may be continuous and controlled by signals emitted by the harvesters.
This requires connectivity between machines, with standardization of formats between different manufacturers. Specifically, between the tractor and the coupled equipment there has been a regulatory package since the 1990s, with the ISO 11783 (Isobus) standard, which is not yet in limited use in Brazil.
There is also a discussion about reducing the size of equipment, which would work in a coordinated manner and with greater agility and accuracy compared to large equipment that dominates the market today. This trend can already be seen in data collection and even spraying operations using drones. These are already autonomous, but based on programmed routes, and despite being a relatively new technology, it is quickly becoming popular.
The biggest limitation to the widespread adoption of this equipment in agriculture is still the costs, with low profitability and, in most cases, seasonal uses of the equipment. Therefore, the development process of these technologies must take into account mainly the demands of each type of crop, the profile of users and regional realities.
Orlando Daniel Masnello and José Paulo Molin, USP/Esalq
Receive the latest agriculture news by email
Receive the latest agriculture news by email
