Agricultural Robot Navigation in Complex Farm and Rugged Terrain Environments Using Deep Reinforcement Learning AI. PhD 36 months PHD Programme By Loughborough University |TopUniversities

Programme overview

Degree

PhD

Study Level

PHD

Study Mode

On Campus

This project aims to use deep learning with a number of sensors to make sense of the complex environment, by detecting and mapping non traversable areas, followed by reinforcement learning for robot navigation. This would provide enhanced robot navigation and control, to avoid areas and obstacles which would normally not be detected and cause the robot to get stuck.
We also aim to use the same Reinforcement Learning techniques in end-to-end control to directly drive the robot’s wheels to enable greater terrain traversable by using a variety of sensor data. These sorts of functions are critical to enabling the sustainable cost-effective use of fully autonomous agricultural robots in agriculture and horticulture, which previously solely rely on GPS and a lot of human intervention to operate. 
Deep Reinforcement Learning is an exciting and extremely powerful AI tool that is seeing a huge increase in use in several areas. It is a machine learning technique that mimics how humans learn by trial and error, reinforcing actions that lead to a goal and ignoring actions that don't. This will enable us to effectively train a robot to operate in these complex environments much faster than using traditional techniques that previously required a robot to have a more semantic understanding of the environment.

Programme overview

Degree

PhD

Study Level

PHD

Study Mode

On Campus

This project aims to use deep learning with a number of sensors to make sense of the complex environment, by detecting and mapping non traversable areas, followed by reinforcement learning for robot navigation. This would provide enhanced robot navigation and control, to avoid areas and obstacles which would normally not be detected and cause the robot to get stuck.
We also aim to use the same Reinforcement Learning techniques in end-to-end control to directly drive the robot’s wheels to enable greater terrain traversable by using a variety of sensor data. These sorts of functions are critical to enabling the sustainable cost-effective use of fully autonomous agricultural robots in agriculture and horticulture, which previously solely rely on GPS and a lot of human intervention to operate. 
Deep Reinforcement Learning is an exciting and extremely powerful AI tool that is seeing a huge increase in use in several areas. It is a machine learning technique that mimics how humans learn by trial and error, reinforcing actions that lead to a goal and ignoring actions that don't. This will enable us to effectively train a robot to operate in these complex environments much faster than using traditional techniques that previously required a robot to have a more semantic understanding of the environment.

Admission Requirements

3.2+
6.5+
92+
A 2:1 undergraduate degree in a relevant subject. An interest or experience with either robotics, machine learning, agriculture or SLAM would be an advantage.

18 Feb 2025
3 Years
Oct

International
28,600

Scholarships

Selecting the right scholarship can be a daunting process. With countless options available, students often find themselves overwhelmed and confused. The decision can be especially stressful for those facing financial constraints or pursuing specific academic or career goals.

To help students navigate this challenging process, we recommend the following articles:

More programmes from the university

PHD Programmes 368