Jianguo Zhao
Colorado State University-Fort Collins, CO·Engineering
How can robots change their shape, stiffness and behavior to cope with difficult environments? The answer could matter for machines that must squeeze through clutter, swim in turbulent water or land on a moving ship.
- Fields
Engineering
Computer Science
- Topics
robotics
shape-morphing structures
artificial muscle
- Grants
- 3 known grants, through Aug 2028
What they work on
- Design modular morphing rods that pair a heat-driven artificial muscle with a variable-stiffness shape memory polymer, plus connectors that let the rods be assembled into reconfigurable 2D and 3D structures.
- Build data-driven models of morphing structures using Koopman operator theory, a way to describe nonlinear motion with linear equations, with deep learning to find variables and graph neural networks to handle large connected assemblies.
- Use the learned models for real-time control and planning that decides which rods to actuate or soften so a structure reaches a desired shape.
- Evolve fish-inspired M3UBot swimming robots by building them through 3D printing and electronic packaging, training them in tank trials with reinforcement learning, and keeping only high-scoring designs across rounds.
- Equip evolved swimmers with hydrodynamic pressure sensors and inertial sensors, then train them to combine basic motions to swim stably in turbulence, follow wakes and tune tail-fin stiffness to save energy.
- Develop drone landing gear with bio-inspired robotic tarsi, insect-foot-like grippers whose compliance is tuned during touchdown, so small quadcopters can perch on surfaces at any orientation.
- Teach drones to start and control perching maneuvers using learned policy regions and vision-based tau-guidance from optical flow, and jointly learn the gear's mechanical design and flight control.
More about the research
Three collaborative projects take up this question from different angles: structures that reshape themselves, fish-like swimming robots designed through a process modeled on natural selection, and small drones that can perch on surfaces at any angle. A common thread is pairing a robot's physical body (soft or tunable materials, compliant parts) with learning-based control. This is sometimes described as combining physical and computational intelligence.
Why it matters
The grants point to concrete uses for these robots. Swimming robots could help with search and rescue, disaster recovery, pollution and ecological monitoring, and underwater infrastructure inspection. Drones that perch on walls, power lines or moving vehicles could extend inspection, surveillance and environmental monitoring, and could land on ships or moving platforms to recharge. Shape-morphing structures could lead to materials that adjust their stiffness, robots that reconfigure to move through complex spaces, and better-fitting wearable devices.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Collaborative Research: MORPHOSIS: Modeling and cOntrol for Robotic and Programmable High-dimensiOnal Shape-morphIng Structures
NSF · Sep 2025 – Aug 2028
Collaborative Research: Artificially-Evolved Modular Robotic Swimmers for Enhanced Mobility in Challenging Hydrodynamic Environments
NSF · Sep 2024 – Aug 2027
Collaborative Research: Omnidirectional Perching on Dynamic Surfaces: Emergence of Robust Behaviors from Joint Learning of Embodied and Motor Control
NSF · Aug 2023 – Jul 2027
Closest labs
Cecilia Huertas Cerdeira
University of Maryland, MD·Engineering·Early-career
Find optimal two-dimensional stiffness patterns for tuna-like caudal fins, for both forward propulsion and maneuvering, by combining soft robotic fins with hardware-in-the-loop optimization.
Suyi Li
Virginia Polytechnic Institute and State University, VA·Engineering
Explore how complex mechano-intelligence (sensing, memory and decision-making carried out by a structure itself) can become when origami acts as a mechanical neural network through physical reservoir computing.
Ioannis Kantaros
Washington University in St Louis, MO·Engineering·Early-career
Train multi-agent reinforcement learning (MARL) control policies for robot teams and mathematically certify their robustness against adversarial, high-dimensional sensor inputs, testing them on aerial and ground robot testbeds and photorealistic simulators.
The school
Colorado State University-Fort Collins: 23 labs on PI Pages, $466M a year on research, 276 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.