Researchers discussing aerial robots around a table in the lab

Robotics · Perception · Systems

We broadly work in algorithms for autonomy with an emphasis on perception, representation and reasoning in robotics. Most of our research is motivated by real-world problems from our field deployments. Several of our projects are in collaboration with experts in related fields as well as domain scientists. Our lab has received support from federal agencies such as NSF, DARPA, ONR, AFOSR, AFRL, REMADE. We are also thankful for support from MOOG Inc.

Applications

All projects
Aerial view of the Erie Canal culvert inspection site

INSPECTOR

Autonomous culvert inspection with legged robots in challenging field conditions.

Excavator arm planning around an obstacle

EARTH

Perception, planning, and hydraulic control for autonomous excavation.

Hexarotor drone conducting field sensing

VISTA

Allocate limited flight time to informative measurements of vegetation condition.

GRAPPLE workflow connecting terrain representations, mobility models, and mission planning

ONR-GRAPPLE

Connect terrain representations and mobility models to robust mission planning.

Representation

GoLF graph representation of curved obstacles and open space

GOLF

Compact spatial graphs that preserve boundaries, bottlenecks, and context for navigation.

HOPHY hierarchy connecting terrain maps to path and mission planning

CLEAR/HOPHY

Reusable terrain abstractions for large-scale off-road path and mission planning.

Perception

Lightning relighting and illumination control pipeline

LGTN

Learn how to control onboard lighting for more reliable robot perception.

NightHawk closed-loop illumination and exposure control

NH

Adapt illumination and camera exposure for visual estimation in dark environments.

Quadruped robot outside Davis Hall with a visualization of reliable visual features

PIXER

Learn which pixels to trust for more reliable, efficient visual navigation.

Reconstructed indoor environment and a plot of its camera trajectory

E3D

Evaluate 3D reconstructions through resolution, accuracy, coverage, and artifacts.

Planning

A safely sampled trajectory passing around obstacles in a narrow maze

SENTINEL

Constraint-aware trajectory sampling for safe, efficient robot motion.

Quadruped robot traversing uneven outdoor terrain

LeggedLocomotion

Connect robot–terrain interaction with locomotion adaptation and risk-aware navigation.

Erie Canal cross-section with the robot and its inspection payload

VISION

Language-guided observation and navigation for autonomous culvert inspection.

Around the lab

Latest news

News archive

August 26, 2026

Charu co-led work on sensing and machine learning to measure recycled plastic content in commercial products.
The technology won a 2026 R&D 100 Award and a Corporate Social Responsibility Special Recognition Award. Read the UB news article.

The people behind the work

Built together

Meet everyone

A place to ask the next question

Let’s build what comes next.

Interested in robotics, perception, or embedded systems?
Get to know our work and connect with the lab.

Join & collaborate