A Multi-tasking AI-assisted Robot System for Individualized Management of Crops and Weeds (AmAlzed)

The use of agrochemicals, including fertilizers and pesticides, has been identified as one of the main causes of biodiversity loss. The EU has set itself the goal of reducing the use and risk of chemical pesticides and reducing the use of more hazardous pesticides by 50%. Existing technologies such as integrated pest management (IPM) and sensor-based application of variable-dose herbicides have the potential to reduce pesticide use by 10-20% without compromising yields. However, progress has been limited and there are even reports of increased pesticide use.

Current weed and pest control practices will not achieve the goal of reducing pesticide use by 50%. Technological innovations for weed and pest control are needed to transform agricultural practices to achieve sustainable use of agrochemicals or even Agriculture without hazardous chemicals without compromising crop yields.

In other words: If pesticides and fertilizers can be applied not only precisely (e.g. by spot spraying and weeding), but also individually (treatments according to the health of individual plants), productivity can be increased in addition to a reduction in these agrochemicals.

Mechanical weed control is increasingly becoming a viable alternative for farmers in Germany. However, traditional mechanical weed control does not kill individual weeds selectively and often leads to soil compaction due to the use of tractors and heavy machinery. Recently, weeding robots have therefore become increasingly available on the market. To date, however, these have rarely had multitasking functions, as they are only integrated into the decision-making process to a limited extent. They are therefore not able to treat a single plant. Using an agricultural robot is far less user-friendly than using a robot vacuum cleaner.

It is still unclear which robotic techniques are more efficient and which data and models are required to treat individual plants and weeds precisely and efficiently. In addition to the questions mentioned above, it must also be assessed whether the new technology has broad application and integration potential compared to existing technologies.

• Simultaneous solution of individual tasks
• low level of AI functions
  
• low working speed due to the inability to complete tasks selectively
• no user-friendliness
• inconvenient transportation
• low reliability
• low programmability: only limited access to all functions and features (main problem in research)

To the above-mentioned problems and the associated scientific scientific questions, experts from the fields of agricultural and bioengineering agricultural and bioengineering, plant cultivation, Horticulture, plant protection and agricultural management from HSWT, TUM and LfL in network. The aim is to develop an AI-supported robotic system with several tasks for the individual cultivation of crops and weeds crops and weeds, with the future vision of fully automated fully automated cultivation of an Agriculture field without the need for labor. labor requirements.

During the two years, the project will focus on the technical on the technical requirements of integrating spectral spectral imaging, environmental sensors, laboratory analyses and AI-based decision decision support tools and to apply the tasks to a programmable ground to a programmable ground robot with a multi-tasking arm that can precisely arm that enables precise treatment of a single plant (e.g. spraying or picking a leaf, pulling weeds out of the soil). weeds from the soil).