Research

UW-Madison MRL focuses on designing and controlling autonomous robots in complex environments, particularly in aquatic domains. We develop both smart “bodies” (e.g., mechanical design, actuators, sensors, communication) and “brains” (e.g., control and perception algorithms) for robots. We teach robots to work as a team to survive in challenging environments and tackle complex tasks that are impossible with individual robots. The lab is also interested in using these robots as tools to explore fundamental questions in aquatic swimming, actuation, sensing, controls, and collective behaviors. We currently have the following research areas:

• Control and Autonomy of Marine Robots
• Bioinspired Aquatic Robotics
• Collective Robotics

Control and Autonomy of Marine Robots

We design autonomous marine robots and develop robust controllers and perception algorithms to empower our marine robots navigating through challenging environments with confidence.

Bioinspired Aquatic Robotics

Marine robots still face challenges with propulsion efficiency, sensing capabilities, and communication. In contrast, aquatic animals have evolved over hundreds of millions of years, demonstrating highly intelligent physical systems. We are creating bioinspired robots and sensors to offer innovative solutions for marine robots, in actuation, control, sensing, and communication.

Collective Robotics

We are developing autonomous robot swarms for aquatic environments using a distributed approach. Our research aims to explore topics such as shape formation, collective transport, and self-assembly. We are also interesting in building bioinspired robot swarms to study collective behaviors in nature.

Research before 2024 (Postdoc and PhD work)

We develop optimization-based controllers, such as model predictive controllers (MPCs) and MPCs integrated with control barrier functions, for marine robots operating in dynamic environments. We are designing learning-based controllers, such as deep reinforcement learning control, to assess the efficacy of these AI models in the robot control.

We seek robust perception algorithms such as SLAM (Simultaneous Localization and Mapping) to enpower marine robots to “see” and understand their surrounding environments.

We are interested in creating autonomouous vessels to address societally impactful problems.

	A fleet of autonomous surface vehicles can be assembled to create on-demand infrastructure, such as bridges, landing platforms, and fences, to facilitate surface activities. We are developing a group of programmable miniature robotic boats that can self-assemble into desired shapes and reconfigure to different shapes on the water in a distributed manner.
	Building autonomous vehicles typically takes a long time, often months or years, relying heavily on human expertise. We are developing an automatic computational design pipeline for jointly optimizing a robot’s shape and controller. We also build real robots to validate and refine our computational design, narrowing the Sim2Real gap.
	We are interested in creating bioinspired aquatic robots and bioinspired sensors. This also motivates basic questions in fish swimming, sensing, and control.
	In nature, weakly electric fish can communicate electrically (eletrocommunication) in the water by generating and receiving electrical signals. Inspired by electrocommunication, we invented a portable artificial electrocommunication system for underwater robots. This artificial communication provides possibilities for affordable solutions in small and micro robots.