AUTONOMOUS CONTROL PLATFORM
Smart farming has introduced a high level of automation and saved millions of tons of pesticides. The missing link for optimizing farming is heading in the direction of autonomous operations in order to optimize resources, increase the level of efficiency and reduce costs. The aim is to transfer technologies from the automotive domain and adapt them to the agriculture sector as much as possible, adding additional features, such as the introduction of new type of sensors, platforms and/or communication networks. The developed V&V tools and demonstrators will improve the testing workflow for autonomous farming systems.
CO-SIMULATION OF AUTONOMOUS FARMING SYSTEMS FOR IMPROVED VERIFICATION AND VALIDATION: This demonstrator provides system developers improved testing scenarios for autonomous farming systems. The systems under test are composed of a combine harvester and a drone demonstrating an autonomous harvesting scenario. The drone has a sensor for creating synthetic hyperspectral camera images and the combine harvester has a sensor for radar simulation. Test scripts allow repeatable and variable test cases and data collection for automated verification and validation. The demonstrator, combined with dynamic features and testing tools, will be a valuable addition for improved verification and validation of the future of autonomous farming, reducing efforts, time and costs.
MODELLING TOOL FOR FLEXIBLE SIMULATION AND TESTING OF FARMING ENVIRONMENTS: The art2kitekt (a2k) demonstrator shows a new web-based toolchain that allows the engineer to perform model-based analysis and simulations. Taking advantage of the flexibility of a2k, a specific scenario of an autonomous farming platform has been modelled acting as a System under Test (SUT). In the demonstrator, both the environment with its static elements and the main agents, the harvester and the drone, have been modelled. The main agents can be decomposed into vehicle dynamics, control, guidance and planner systems. All these components are configurable and interchangeable, and can be individually tested using the provided checkers. It is also possible to execute the simulation in “live” or “fast” mode to reduce the verification time. Inspired by the standard FMI, a2k allows the flexible simulation of different components, its interconnections, checking the outputs and visualizing the results via viewers, tables, etc. Additionally, the a2k framework, through its Quality Management component, allows the engineer to easily generate test plans and test cases.
They can be executed in “live” or “fast” mode and then show the results. This is the first web-based tool chain that allows the simulation and testing of farming environments.
RUNTIME VERIFICATION AND AUTOMATED TESTING OF IN-VEHICLE COMMUNICATION: The farming machines in the Automated Farming Use Case are envisioned to use Deterministic Ethernet (IEEE 802.1 TSN or TTEthernet) for in-vehicle communication, to be able to integrate high-bandwidth data from the sensors, other vehicles (including drones) and infrastructure. The demonstrator will showcase:
- the TTEthernet Automatic Test Framework from TTTech which runs applications and generates, monitors and analyzes TTEther-net traffic over the network,
- the TTCONTROLonf tool from TECHNICAL UNIVERSITY OF DENMARK which provides a timing analysis to test the timing properties of the applications and
- the RMTLD3Synth online tool that is responsible for generating correct-by-construction runtime monitor code for observing real-time traces and reason about the correctness of duration properties of events within those traces.
Automatic testing of Deterministic Ethernet communication, (2) Network Calculus-based timing analysis that can support both TSN and TTEthernet and (3) generation of runtime monitors from formal specifications, in a correct-by-construction manner and that are able to identify real-time properties that consider the duration of events.
The challenge of this Use Case is to validate a new concept for autonomous farming by removing the human user from the cabin of the farming vehicles and allowing him to act as a remote supervisor of the complete farming system. With that in mind, the aim of the Use Case is to evaluate a plethora of technologies, arising from distinct areas, and see their applicability and usability in the very promising field of autonomous farming and work towards an integrated V&V demonstrator.
The several technologies being considered in the Use Case include state-of-the-art work on: an hyperspectral flying platform consisting of a UAV equipped with an onboard hyperspectral camera, software, algorithms, and communication infrastructure; a virtual farming environment, where the farming machines operate (a combination of an harvester, a tractor and drone) and can be controlled by real control systems or embedded algorithms; environment stimulation models to test radar sensors which represents the real world radar sensor with the purpose of creating the possibility of analyzing the radar sensor behavior in virtual environments; simulation software where a 4x4 drive vehicle configuration is used to have better traction grip and thus be applicable to the characteristics of a tractor; a simulation platform for enabling the V&V of different hardware components which allows to be connected to the real environment via the interfaces defined for the drone and/or the other vehicles; and, finally, an automatic testing framework for intra-vehicle network communication consisting of a combination of highly advanced network simulation tool and mechanisms to verify worst-case transmission delays and the generation of runtime monitors to verify the satisfaction of those delays during execution time.
The Use Case team is currently working towards the envisioned autonomous farming V&V concept by putting efforts on the integration of the several technologies described. When that is achieved, the resulting verification framework will improve the testing workflow for autonomous farming systems and pave the way for upcoming work with respect to this domain.