The world of autonomous agriculture is hitting a turning point, no longer are autonomous machines confined to research and proof of concept but entering the world of production, and this brings multiple challenges for manufacturers especially those wanting to scale and create full production quality machines.
As the standards for autonomous agriculture such as ISO25119 and ISO18497 become more widely understood and applicable to the industry, there are certain processes and procedures that need to be followed to reduce the safety and operational risks associated with designing and operating autonomous agricultural machines. Beyond the operation of the autonomous platform, there are two key safety areas which are critically important to ensure the safe behaviour of the machine. The first is the geofence, an area where the machine can operate. The second is the perception system, the system used by the machine to detect if an object or obstacle may be in the path of the vehicle.
This post focuses on positioning and how it is critical for safe operation in agriculture robotics. Starting with the requirements of a geofence, it is clearly stated within the above-mentioned ISO standards that the machine must not operate outside of a defined working area. The defined working area may be the boundary of the field, and it may include fixed objects within the field, such as bodies of water, electricity poles and trees. However, consideration needs to be made as to the actual position of this boundary and fixed objects and how the position is obtained in the first instance through prior mapping.
Boundary logging security
Most field boundaries today, are mapped with a GNSS receiver. While these boundaries are typically good enough for most farming practices, when it comes to autonomous machines, there can be significant errors if not done with precision and accuracy. If the correction source and position of the antenna in relation to the boundary are unknown the integrity of the boundary will not be accurate enough for autonomous machine operations.
For example, if a field boundary was captured with an SBAS correction source, the position may drift over time. This means that the position of the boundary can also drift and begin to include areas that were originally excluded by the geofence.
The GNSS receiver manufacturer and the correction source provider also need to be on the same datum to ensure the physical and digital worlds align repeatedly year after year safeguarding the machines geofence integrity.
Actual geofence location
Once an outer boundary of a geofence is captured, and the digital and physical boundaries match, an exclusion zone around the boundary needs to be mapped. This provides a safety zone for when a vehicle begins to approach the inner boundary, providing enough space for the machine to slow down, turn or stop, ensuring it does not cross the physical boundary.
While the speed and width of the machine are fixed at the start of an operation, the heading is variable. To prevent crossing of the physical boundary the autonomous machine must always have a known heading. In a single antenna solution, the heading becomes inaccurate at slow/stationary speeds, as movement is required to accurately compute the GNSS-derived course of the machine, while a dual-antenna solution can provide an accurate heading while operating at slower speeds.
Hexagon | NovAtel’s ALIGN technology computes a high-accuracy heading solution with two GNSS SMART antennas, providing the vehicle’s true physical heading at all times, even when it is stationary or operating at low speeds.
GNSS and position uptime
As you can see, GNSS technology is critical to the operations of autonomous machines. It provides position, velocity, heading and timing information, which is integral to precision farming practices and is vital in enabling a geofence for agriculture robotics.
If GNSS signals are degraded, partially blocked, or otherwise not able to be received, SPAN technology from NovAtel offers the ability to bridge a temporary loss of GNSS signals to ensure continuous, reliable positioning. In standard agriculture equipment, operated by humans, this means that the operator can continue to follow their line without the auto-steer cutting out or having large position jumps, enabling advanced functions such as automatic end of row turns and accurate boundary along treelines. For agriculture robotics, this is important to keep the machine operating so it is not stopped in the field unnecessarily allowing customers and OEMS to get the highest uptime from their autonomous deployment.
As GNSS technology relies on signals provided from space, there will always be the risk of interruptions to those signals with atmospheric or local interference, or to the externally provided correction source. Similarly, even with a high-quality hardware installation, there is always a small chance of hardware and/or software-related issues.
NovAtel can help reduce these risks which can result in degradation or, in the worst case, complete loss of positioning. Our portfolio offers multiple corrections services, positioning engines, hardware and software profiles, for any machine manufacturer to be able to keep their operations moving. Integrating these technologies into the system architecture can reduce the risks identified by the manufacturer’s Hazard and Risk analysis and meeting their safety goals.
Integration of reliable and robust GNSS solutions into autonomous machines for agriculture requires a new way of thinking. The heading and orientation must always be known for geofencing, to maintain position integrity and ensure machines never breach physical boundaries. NovAtel is at the forefront of GNSS technology and has a full selection of robust and reliable solutions that can advance agriculture robots in achieving compliance with the relevant ISO standards.
Our customizable hardware and software solutions support machine manufacturers in meeting their functional safety goals and are ready to scale from development to production for the next generation of farm machinery.