Modern agriculture is increasingly adopting AI-enhanced autonomous systems to enable more precise, adaptive and sustainable crop management. In this talk, drawing on a series of research projects, I will show how UAVs equipped with multispectral sensors and machine learning models have been deployed for early detection of crop diseases (e.g., wheat yellow rust), pest pressure and water stress. I will then discuss our recent work on crop spraying drones, including spraying distribution modelling and optimal path planning, aimed at reducing chemical use while maintaining efficacy in crop protection.
Although these systems are currently modular and task-specific, I will outline a conceptual framework for closing the feedback loop, from sensing to inference to intervention, enabling future autonomous systems to act adaptively in crop management. The talk will conclude with open challenges and opportunities in achieving truly autonomous and adaptive agricultural systems.
Agriculture faces unprecedented challenges, ranging from labour shortages and rising productivity demands to the imperative for sustainable practices. This presentation delves into the specific attributes and difficulties within the agricultural sector. By showcasing Monash University's progress in robotics and AI, it illustrates how these innovative technologies are transforming agriculture and contributing to a sustainable future.
Plants have evolved extraordinary strategies to adapt, grow, and thrive in dynamic and often harsh environments, making them a powerful source of inspiration for robotics in complex natural settings. Leveraging these principles, we present a new class of plant-inspired robotic systems tailored for precision agriculture—autonomous, multifunctional machines capable of navigating unstructured terrains, monitoring microclimates, and interacting delicately with living crops. These bioinspired and biohybrid devices integrate the morphological and biomechanical traits of selected plants, allowing them to precisely adapt and interact with the terrestrial and/or aquatic environments. Fabricated using advanced techniques such as microcomputed tomography, two-photon lithography, and bioprinting, these systems are designed for sustainability, scalability, and in-field resilience. Tested in real-world environments (such as soil, leaf tissues, and aquatic habitats) these machines have demonstrated their potential in applications like climbing robots, precision agriculture, reforestation, and underwater sensing. By merging plant-inspired design with robotics and microsystems engineering, these technologies open new frontiers in sustainable agriculture, offering enhanced precision, reduced resource waste, and improved adaptability in next-generation farming systems.
Agriculture is a challenging environment requiring novel and unconventional solutions, for both the mechanical design and control and learning. In this talk I will introduce different robotic technologies, including soft robots, modular ones and cable driven, and approaches for deploying robots in the field. I will also touch on unconventional problems or focuses within agriculture, and how robotics and control can be applied.
In this talk, I will report on our recent progress in developing algorithms and systems to perform robotic weed removal tasks in precision agriculture. Weed removal is an eternal issue in agriculture. However, the task is often labor intensive or has a large environmental impact if herbicide is used. Using robots can address those issues and make it possible to provide an environmentally friendly approach in weed management. Here, we present three case studies. In the first case, we developed a new deep learning method to effectively distinguish nutsedge weeds from background turf grass. The challenge is to reduce human data labeling costs by designing a new network and features and combining synthetic data that enable network training with a small amount of data with low labeling effort. In the second case, we will discuss how we design a robotic micro-volume weed spraying system and motion planning algorithms to suppress weeds in early growth stage. If time permits, I will discuss our latest progress in weed flaming by employing a mobile manipulator with a SPOT mini quadruped robot and a 6-degree of freedom (DoF) robot.