MAE Triton AI
Electric Autonomous Go-Kart
Department of Mechanical and Aerospace Engineering
University of California San Diego
Sponsor:
Dr. Jack Silberman
Team:
Adrian Castaneda, Jack Griffin, Hanson Nguyen, David Qiao
Electric Autonomous Go-Kart
Department of Mechanical and Aerospace Engineering
University of California San Diego
Sponsor:
Dr. Jack Silberman
Team:
Adrian Castaneda, Jack Griffin, Hanson Nguyen, David Qiao
Triton AI is a student organization at UCSD, led by academic advisor Dr. Jack Silberman, that focuses on learning artificial intelligence and applying that knowledge towards competing in various robot competitions. Previous projects include the Jetracer, a small-scale autonomous racing vehicle, and the Triton Racer, a computer vision based autonomous driving platform for robotic cars. Currently, Triton AI's main project is building an electric autonomous go-kart to compete at the evGrandPrix Autonomous racing event.
The evGrandPrix Autonomous is an autonomous racing event for students in which universities develop their own software, powertrain, steering, and throttle systems and compete against each other. The event is hosted in Indianapolis at Purdue University's Grand Prix Track and takes place on May 20, 2022. This is just the 2nd time that the event is being held fully autonomous. This year's race consists of a single-kart time trial run of three laps in which the winner of the event is the kart that has the fastest lap time.
Purdue Grand Prix Track
The go-kart's chassis was provided by Top Kart USA, which is uniform for all the other teams competing in the evGrandPrix. However, it was our team's job to improve upon the kart's previous steering, drive motor, and braking systems in order to ensure that all of the kart's mechanical systems were reliable for the race. This was done by improving the structural integrity of the kart, implementing new designs, and performing many failure tests. Additionally, we worked in parallel with members of Triton AI and the ECE 191 capstone team who worked on the software and electrical components of the kart. Due to the Dr. Silberman's desire to have an affordable, easily replicable, and open-source design for other universities, it was important to document the kart's bill of materials (BOM), which includes any parts that were ordered online or fabricated in the school's machine shop. A summary of the project's objectives can be seen below.
Project Objectives
Ensure reliable operation of steering, drive motor, and braking systems for the evGrandPrix on May 20, 2022.
Suggest possible improvements to the steering, braking, and drive motor systems that can improve predicted simulation lap times.
Implement/propose a mechanism to allow for human override of autonomous control to be used by Triton AI to collect data for future improvements.
Final Design
Using the Top Kart USA chassis as a starting point, various mechanical and electrical components were added to the kart throughout. The back of the kart includes delrin (acetal) shelves to hold the electrical components while the front of the kart contains aluminum T-slotted rails to support the steering and camera systems. The right side of the kart holds the drive motor system while the left side includes the braking system.
The steering system includes a NEO BLDC steering motor and a MAXPlanetary gearbox kit with a 100:1 gear ratio. The steering motor shaft is a 1/2" diameter shaft that is hexagonal on one end to fit in the motor and circular on the other end to fit in the bearing and bushing. The circular end includes a key slot to prevent slipping of the bushing. A pulley and timing belt fits around the bushing and connects the steering motor shaft to the steering wheel shaft.
1/2" diameter shaft that is hexagonal and circular. The key insert prevents the bushing from slipping.
The motor rotates the steering motor shaft which rotates the steering wheel shaft and turns the tires using the pulleys and timing belt.
Adjusts belt to optimal tension to prevent the belt from skipping teeth on the pulley while steering.
The driving system for our kart uses a great-performing ME1719 brushless DC motor with a powerful electronic speed control, the VESC 75/300. The speed of the kart can be easily controlled using the VESC's built-in PID speed control functionality. Our drive system has a pulley ratio of 2.5, which is optimized for general kart racing at around 30 mph.
The braking system includes a linear actuator that extends and pushes on a brake lever, which supplies pressure through the brake wires to the brake caliper. The brake caliper acts to slow down the rear wheels of the kart. In addition, the ME1719 drive motor provides electronic braking that is very reliable at lower speeds.
2022 evGrandPrix Race Results
UCSD finished in 2nd place (behind Kennesaw) with a constant speed of 3 m/s and a lap time of 2 minutes and 45 seconds, and was one of two teams that were able to complete a lap fully autonomous. The kart was able to complete laps without any mechanical failures and had reliable operation of steering, drive motor, and braking systems. Software GPS navigation issues, in which the mapping of the track was slightly offset, caused the kart to steer slightly off the track and prevented the kart from driving at higher speeds and achieving a faster lap time.
In the 2021 evGrandPrix, UCSD collaborated with University of Hawaii and used their kart for the event. However, they did not have a functioning steering system and could only drive forward in straight segments, which prevented them from completing a lap.
MAE Triton AI Team with 2nd place Trophy after the Race
Video of Kart in Action