Simplifying autonomous navigation of agricultural robots
Abstract
As the global population continues to grow and is projected to reach around 9 billion by 2050, the demand for food is escalating at an unprecedented rate. This surge in population necessitates a corresponding increase in food production to ensure food security and prevent hunger. The importance of increasing food production cannot be overstated in the context of a growing global population, economic development, and environmental sustainability.
Agricultural automation is imperative to increase food production. The advent of autonomous agricultural robots is revolutionizing farming practices and significantly contributing to increased food production. These robots utilize advanced technologies such as artificial intelligence, machine learning, and the Internet of Things (IoT) to perform various agricultural tasks with high precision and efficiency. By integrating cutting-edge technologies into agriculture, these robots help maximize yields, optimize resource use, and build resilience against climate change, ensuring a stable and abundant food supply for the future.
Navigation is an essential part of autonomous agricultural robots. Besides Global Positioning System (GPS) based methods, most autonomous navigation methods are complex and require high computational power, an energy burden for small agricultural robots. Power issues are not critical for big autonomous agricultural machinery like planters, harvesters, or balers. Our concern is for row crop navigation of small agricultural robots, where running a high computing load for a long time is a real problem. Another concern is the complexity of hardware and software because high energy demands come from cutting-edge hardware and complex algorithms, which are sometimes inconvenient or challenging to implement in the real-time operation of robots.
This dissertation aims to find ways to make autonomous navigation of robots easier by creating a simple pipeline. It is divided into three main components: mapping, sensing, and control to facilitate autonomous row crop navigation that is needed for the efficient application of agricultural robots.
The first chapter introduces the importance of robotic mapping, its types, and how Unmanned Aerial Vehicles (UAVs) can be used to map navigable areas (rows) between crop rows. Estimating the navigable area is vital for energy budgeting, optimizing routes, and various field operations. The second chapter explored flex sensors as simple tactile sensors that can be used to enhance the autonomous navigation capabilities of agricultural robots. The chapter discussed the necessity of tactile sensing in agricultural scenarios. We showed that flex sensors, which detect changes in resistance caused by bending or flexing, can be used to distinguish between plant parts like leaves and stems. Sensing this difference allows agricultural robots to pass through leaves or soft stems. In this way, flex sensors help to assess any navigable areas as passable or non-passable. The third chapter presents the application of fuzzy logic with tactile sensors (flex sensors) for obstacle assessment by autonomous robots in agricultural scenarios. Based on the evaluation, the fuzzy control can direct the robot to pass through or avoid obstacles. The output from the flex sensors is used to drive a fuzzy logic controller that controls the robot's wheel movement to navigate through crop fields or forest areas autonomously.
In addition to the above three studies in three chapters, there are two more chapters: one chapter sheds light on the importance of a dedicated simulator for agricultural robotics, and the other chapter describes point clouds' technology with its future applications for agricultural automation. These two studies also aim to show ways to simplify and efficiently automate navigation tasks by farming robots.
In summary, this study investigates ways to make autonomous navigation of agricultural robots simple and efficient with standard tools and established methods. It will help make autonomous navigation of agricultural robots through row crops easier and inspire others to think in simple ways to tackle challenges of complex tasks using conventional tools.
