Development and Integration of a Computing Platform for an Autonomous Vehicle - James Wyatt, 2018
Monash Motorsport Final Year Thesis Collection
The Final Year Thesis, is a technical engineering assignment undertaken by students of Monash University. Monash Motorsport team members often choose to conduct this assignment in conjunction with the team.
These theses have been the cornerstone for much of the team’s success. The purpose of the team releasing the Monash Motorsport Final Year Thesis Collection is to share knowledge and foster progress in the Formula Student and Formula-SAE community.
We ask that you please do not contact the authors or supervisors directly, instead for any related questions please email info@monashmotorsport.com
Summary:
What follows in this report is the design process of an autonomous racing vehicle’s computational system and structure. This covers the original research performed, the concept generation phase, and the final design decisions. It concludes with detailed documentation of the system architecture, and various tasks that were performed in tandem.
The system is designed with safety in mind, and utilises a chain of heartbeats to ensure that all processing nodes are alive and functioning correctly. The last line of defence is built using non-programmable physical logic gates which are unable to fail due to poor programming. This computing system was used in 2019 as the foundation for Monash Motorsport’s first autonomous vehicle, M19-D.
Introduction:
At the beginning of 2018, the autonomous management team decided upon a single goal:
To build an autonomous vehicle capable of completing all competition events in 2019.
This goal underpins everything that the autonomous subsection has done and shall continue to do into the future. Whilst some teams choose to optimise their vehicles for specific events, Monash Motorsport aims to be the most respected FSAE team in the world, and performing well in just one event is not enough to achieve this goal. At the beginning of 2018, the autonomous team was split into subsections, these include
LV Systems
Cameras
LiDAR
GPS/INS
Computing
Path Planning
Vehicle Actuation
Software Infrastructure
Specifically, in the case of computing, the following requirements were generated at the beginning of the year:
Provide a unified platform for other subsystems to use for processing.
Provide a method to share information between multiple subsystems
Interface with the existing ECU, or, implement the functionality of the existing ECU
Be able to process images and feed-forward pre-trained neural networks
Implement a state-machine to keep track of the vehicle status
Conclusion
These requirements have been satisfied:
The hardware and utilities to interface with the hardware have been created. All of the computing components are linked, either through Ethernet or UART, and there exists a mechanism to interface with the vehicle.
Robot Operating System provides a robust method to share information between subsystems.
Through the PSoC 5LP, the system can interact with the ECU over CAN.
The Jetson TX2 is capable of processing images and feed-forward neural networks
networks.
The PSoC implements the state machine.
On top of these initial requirements, new issues have come to light that needed to be solved. Safety of the autonomous system was not originally considered, however through the course of development it was realised that safety is actually the number one concern. The method implemented, using a combination of heartbeats and physical hardware solves this issue elegantly and robustly.