Farm Field Coverage Path Planning for Autonomous DOT Vehicles using Reeb Graphs and Reinforced Learning

The objective of this project is to develop full-coverage path plans for an autonomous vehicle, designed and developed by DOT Technology Corp., operating in a farm field. Conventional methods used in the realm of computational geometry, such as: (a) converting a digital representation of the physical space, i.e., the farm field of interest, to a Configuration space (C-space), where the robot can be seen as a automaton, i.e., a point robot, (b) breaking the C-space representation of the field to cells, via conventional cell decomposition methods, (c) generating the adjacency maps on decomposed cells, and (d) finding the optimal path for each cell that yields the best coverage, lowest number of turns, and minimal overlaps between runs will be investigated.

Faculty Supervisor:

Mehran Mehrandezh

Student:

Behnam Nasirian

Partner:

DOT Technology Corp.

Discipline:

Engineering

Sector:

Manufacturing

University:

University of Regina