Toward Intelligent Adaptable Robots for Agriculture – Tiara

The use of robotics in agriculture appears as a lever capable of supporting the necessary reduction of the environmental impact of human activities, while maintaining production levels compatible with a growing world population. The necessary reduction of chemical products indeed requires the development of new tools for agriculture, allowing to increase the frequency of treatment without requiring an increased manpower. In addition, the arduousness and the repeatability of agricultural activities brings a difficulty of recruitment. Such a lack of manpower can be offset by the deployment of machines with a high level of autonomy, reducing the hardness of the work. As a result, more and more robotic solutions are emerging in the field of agriculture, making agricultural robotics an growing industry. Nevertheless, the degree of maturity still limited requires human supervision, sometimes depreciating the efficiency of such robots. As a result, the development of new agricultural machines, whose level of autonomy can be adjusted, and using an energy source that respects the environment is a valuable asset in the context of the transformation of agriculture. The Sabi-Agri company was created in 2007, in such a context. It aims at proposing new electric, modular and agile agricultural machines, currently controlled manually but easily compatible with automation.
The joint laboratory aims to develop the autonomy of this type of agricultural vehicles. It proposes to implement several robotic behaviors, as well as adaptation mechanisms to select and adjust these behaviors according to the task to be performed and the context of evolution. The main objective is to develop a versatile control architecture, allowing the vehicle to work autonomously in interaction with humans. Tiara will lead in particular research on decision making processes (IA) for the on-line reconfiguration of basic robotic behavior. Indeed, the diversity of the situations encountered, as well as the variability of the interaction conditions between the robot, its implement and the environment, requires the use of several robotic approaches, to allow the robot to guarantee the safety and precision of the work achieved. This research will have to account for human actions, both with respect to the decision-making processes, but also in real time. To this end, the control architecture will integrate aspects of collaboration with the human (cobotics), with particular attention to the dialogue between Man and robot.
The constitution of the joint laboratory is considered in a progressive way, by first putting in place several elementary behaviors already tested on other types of robots (tracking of multi-sensor trajectory, pursuit of vehicle or pedestrian, ...) , in order to bring a first level of assistance to the vehicle within 1 year. This first step will then provide a hardware and software architecture for the development of decision-making processes integrating human behavior, providing a high level of autonomy after 2 years and capacity for reconfiguration and autonomous decision after 4 years. years. This vision will also allow the aggregation of additional behaviors in year 2 and 3, according to the needs in the field, as well as the automation of embedded tools, through generic research conducted on mobile manipulation in natural environments. Given the maturity of autonomous navigation approaches, this last component is the culmination of robotics in agriculture, in order to concretely implement a precision farming, capable of considerably reducing the impact of human activity on the environment.