Robot Open Autonomous Racing (ROAR™)

Led by its faculty members with deep expertise in AI and autonomous driving, Berkeley is proud to announce and host a new AI racecar competition in 2020. The Robot Open Autonomous Racing (ROAR) competition will pit multiple student racing teams to compete for speed and vehicle skills at the heart of the iconic Berkeley campus.

The participating teams will adopt the set of common 1/10 RC car hardware as regulated by a rules committee. The regulations aim to ensure the competition to be affordable and fair to our target students. Different teams are allowed to upgrade their car hardware as permitted by the regulation and to develop and deploy their own autonomous driving software.

In order to preserve the safety and well-being of all students and members in our community, the decision has been made to host a virtual debut competition for ROAR in Fall 2020.

Participation:

Submission website for ROAR 2021 Spring Season will open in April. You need to register to be a member of the Vive Center to be able to submit.

Steering Committee:

Francesco Borrelli

Shankar Sastry

Koushil Sreenath

Allen Yang, Chair

Leagues:

* Important Note: Due to restrictions imposed during the COVID-19 pandemic, ROAR competitions for 2020 and 2021 seasons will only host the S1 series until further notice.

S1 Series (Software Simulation)

V1 Series (Virtual Reality)

A1 Series (Autonomous Racing)

S1 Series (Software Simulation)

ROAR provides a Python-based racing simulation environment that allows contenders to race their autonomous AI agents. Contenders can fully train and test their AI algorithms without any vehicle hardware.

For more detail, please see: https://github.com/augcog/ROAR

V1 Series (Virtual Reality)

The ROAR reference hardware and software uniquely supports contenders to manually race their vehicles in virtual reality. The competition will pit different players to demonstrate their racing skills in 1/10 RC car style.

For more detail, please see: https://github.com/augcog/ROAR_VR

A1 Series (Autonomous Racing)

A1 series is the pinnacle of the ROAR program, where user-developed AI agents will need to fully and autonomously control a RC vehicle to complete 10 laps of the challenging physical race track as in the V1 series. The ROAR program provides a reference design of the ROAR vehicle system that allows seamless deployment of autonomous racing agents between Software Simulation (S1) and Autonomous Racing (A1)

For more detail, please see: https://github.com/augcog/ROAR_Jetson

Fall 2020 Competition Results:

For the Fall 2020 season, we have received 8 submissions by December 12. These are the final results in terms of the eight team’s total time for 10 laps and the best single lap time.

Team Members	Total Lap Time	Best Single Lap Time	Remarks
Alfredo De Goyeneche, Alvin Tan, Sihao Chen, Wesley Wang, Aman Sidhant	813.92 s	77.23s	Using customized waypoint smoothing function & LQR Controller
Shuwen Deng, Zheyuan Wu, Linfeng Zang	1104.00s	77.15s	Fast driving using manually-tuned PID controller
Michael Wu, Flaviano Christian Reyes, James Cheney, Marleah Puckett, Jonathan Wong	1207.45s	120.45s	Using Reinforcement Learning DDPG network trained PID controller
Josephine Koe	1294.13s	126.96s	Using optimally tuned PID controller with customized local planner
Peter Peng, Amanda Chiang, William Situ	2531.55	253.10s	Driving at 44 km/h. Safe but not fast
Jacky Kwok	DNF	---
Kevin Liu	DNF	---

We are very pleased to announce the winners for the 2020 ROAR S1 series:

Grand Prize: Alfredo De Goyeneche, Alvin Tan, Sihao Chen, Wesley Wang, Aman Sidhant (Record: 813.92s)

Second Place: Shuwen Deng, Zheyuan Wu, Linfeng Zang

Third Place: Michael Wu, Flaviano Christian Reyes, James Cheney, Marleah Puckett, Jonathan Wong

Special Awards:

Fastest Single Lap: Shuwen Deng, Zheyuan Wu, Linfeng Zang (Record: 77.15s)

Prime Directive (Fastest system that had zero crashes during competition): Michael Wu, Flaviano Christian Reyes, James Cheney, Marleah Puckett, Jonathan Wong (Record: 1207.45s)

Courses:

Courses offered at UC Berkeley in Fall 2020 where students can learn more about ROAR.