Embedded Files
Triton AI is a student-run robotics organization at UC San Diego founded in 2019 and led by Dr. Jack Silberman that focuses on applying machine learning and artificial intelligence to robotic vehicles to race in competitions.
EvGo-Kart
EvGo-Kart
The robot is a race-grade electric go-kart (evGoKart) that competes in autonomous evGoKart races such as the evGrandPrix autonomous and the Self Racing Cars competition. The chassis was provided by Top Kart USA.
The Triton-AI evKart previously won 3rd place in 2021 and 2nd place in 2022 in the evGrandPrix autonomous race. The team is now shooting for 1st place in 2023.
Project Overview
Project Overview
Our objective is to design and test the subsystem for the autonomous go-kart to steer using computer control while safely allowing for human override at any moment in order to create a development platform for vehicles that are human/autonomous driven. Dr. Silberman has the desire for this project to be affordable and easily replicable for other colleges to follow.
Functional Requirements
Functional Requirements
Torque capacity > 50 Nm (the measured torque required to steer the vehicle)
Measure the angle of the steering wheel
Electronic power steering for autonomous driving mode and assisted steering for human driving
Allow manual human steering during a loss of power
Measure torque applied onto the steering wheel by human
Differentiate between human and motor input
finalvid.mp4Final Product Movie
Final Design
Final Design
The steering actuator is an electric power assisted steering (EPAS) system from a 2017 Toyota Corolla. Reasons this unit was chosen included its torque output that exceeds that required by the go-kart, the built-in torque sensor that provides the ability to detect human input, its extensive availability/affordability on eBay and the possibility to hack into its controller are network (CAN) interface.
CAD of Final Steering Design Assembly
Performance
Performance
The 2017 Toyota Corolla electric power assisted steering (EPAS) system was able to be temporarily hacked and controlled through its CAN interface. After subsequent failures with the CAN, control of the unit was passed to a Vedder's Electronic Speed Controller (VESC). An external encoder is paired with the VESC to track the steering angle. With this feedback-loop closed the EPAS is capable of rotating the steering wheel to the desired steering angle. Furthermore, the EPAS has highly responsive feedback of the torque input that can be determined by differentiating the voltages from its torque sensor. This system fulfills the current requirements and provides the ability for future development of a human interface.
Page updated
Report abuse