MUBot
AI-guided robotic fish with flow sensing and modular buoyancy control for noninvasive, low-power underwater monitoring
The follower MUBot learns to track the leader's location autonomously using only flow pressure, without vision or contact, trained through imitation learning.
Conventional underwater sensing relies on active sonar/lidar or vision—approaches that add bulk, draw power, can fail in dark or turbid water, and may disrupt wildlife. MUBot, a modular robotic fish developed by researchers at Penn State, aims to address these limits with two capabilities: (1) a lateral-line–inspired pressure sensor array that reads ambient flow; using imitation learning, a follower µBot tracks a leader purely from pressure cues—no vision or contact—enabling short-range, covert monitoring even in darkness; and (2) 3D-printed, per-module buoyancy control devices that tune local density, paired with a fast-actuating dorsal fin. Multiple buoyancy modules allow switching among positive, neutral, and negative buoyancy and provide pitch control for agile vertical maneuvers in a small, flexible body. Designed to be easily extended with camera modules and other environmental sensors, MUBot has the potential for autonomous monitoring and station-keeping in the wakes of rocks and other structures in rivers—regions rich in aquatic life. By combining passive, non-emissive flow sensing with fine buoyancy control, MUBot points toward compact, fieldable robotic fish that can quietly track and study aquatic species and habitats with minimal power and disturbance.
Pressure sensors and buoyancy control devices integrated across MUBot generations, with custom mechanisms and electronics.
Learn more about the MUBot here: https://youtu.be/W5Z2T9xcR3k?si=3jpRIS8QPJd0GKaI
Panta, K., Deng, H., DeLattre, M., & Cheng, B. (2025). Leader-Follower Formation Enabled by Pressure Sensing in Free-Swimming Undulatory Robotic Fish. 2025 IEEE International Conference on Robotics and Automation (ICRA), 14685–14691. https://doi.org/10.1109/ICRA55743.2025.11127903