Tardigrade is our latest autonomous underwater vehicle, designed for the 2025 RoboSub competition.
After a couple years of setbacks, Tardigrade represents a fresh start with a streamlined design that
prioritizes consistency and versatility. The vehicle features dedicated electrical housings, a central housing,
integrated sensors, and robust control systems that enable precise maneuverability and reliable task execution.

• Latest generation AUV
• Compact, modular frame
• Integrated sensors and housings
• Robust control systems
• Optimized for RoboSub 2025 tasks

Currently in development, our 2021 RoboSub AUV has been designed to be 33% smaller and 40% lighter than our previous AUV. Please feel free to follow our major updates by checking out our Facebook page, or find us on Github for a closer peek at our daily progress.

• 33% Smaller
• 40% Lighter
• Most computation power
• Multimodal Solutions

Bearracuda is UR@B’s first competition ROV that will be competing in the next International MATE ROV Competition! This vehicle embodies the resilience that our team has demonstrated throughout the Fall 2020 semester through navigating the entire design process virtually.

Bearracuda features a carbon fiber frame and multiple custom payload tools to complete missions regarding plastic pollution, climate change, and coral reefs. Specifically, Bearracuda is able to retrieve submerged objects (nets, bottles, coral samples, etc), conduct environmental studies using image recognition, and deploy devices/sensors for measurements and sampling.

• Our first ROV for MATE ROV competition
• Able to retrieve submerged objects
• Image recognition
• Environmental sensors for measurements and sampling

Blue is the first autonomous underwater vehicle that Underwater Robotics at Berkeley built (back when we were known as AUV’s at Berkeley). Blue features a stereoscopic vision system, data cataloging abilities, and a robotic arm. Blue is designed to complete image recognition tasks and manipulate objects with a robotic arm. Funding for Blue was largely made possible by OmniSci, whose software was used.

• First generation
• Steroscopic vision system
• Image recognition task ready
• Data cataloging

Smol Bot was the first operational submarine robot built as a test bed for controls, machine learning, and computer vision teams. The robot has tethered power cables and a 15ft USB chord so software teams can modify and deploy code while the vehicle is underwater. This enables the autonomous robot to be manually controlled and allows the team to execute different tasks in the same underwater dive. Funding for Smol Bot was made possible through grants from the UC Berkeley College of Engineering and Engineering Student Council.

• Test bed for controls
• For machine learning
• For Computer vision
• Funded through ESC