Off-Road Autonomous Path Following in an Instrumented Small-Scale Test Vehicle

Open Access
- Author:
- De Lattre, Micah
- Area of Honors:
- Mechanical Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Sean N Brennan, Thesis Supervisor
Bo Cheng, Thesis Honors Advisor - Keywords:
- Path-following
autonomous vehicle
off-road
Simulink - Abstract:
- This thesis investigates high-speed off-road autonomous path-following using an instrumented small-scale test vehicle. The proposed test vehicle employs a combination of sensors, including GPS and rotary encoders to accurately track the vehicle's position, orientation, and speed during traversals of an off-road course. An onboard microcontroller collects and stores this sensor data in real-time which can be post-processed later. Steering and throttle limiting algorithms are implemented onto the vehicles onboard microcontroller to enable steering and throttle control. A Simulink trajectory following model was utilized in this work to simulate the vehicles traversal of a provided trajectory. The results of the simulation allow algorithm selection and controller tuning to provide the necessary steering angles and velocities to enable path following. At Penn State’s test track, an off-road area was chosen as the test course for off-road driving experiments in future work. This test course was designed to feature numerous switchbacks, a straight-away where maximum speed can be achieved, steep inclined hills, rapid changes of elevation, and varying terrain. These features increase the complexity and detail needed in a proper path-following algorithm. This research discussion concludes with discussion of next steps for field testing. The overall goals are to contribute to the development of autonomous off-road vehicles for use in a range of applications, including agriculture, mining, and search and rescue. The findings of this thesis have implications for the advancement of autonomous systems technology in challenging off-road environments.