Research

Advancing the science of autonomous construction

Our research spans machine learning, computer vision, robotics, and systems engineering. We publish our findings and contribute to the broader scientific community.

Neural Frameworks

We develop deep learning architectures optimized for real-time decision making in dynamic construction environments. Our models are designed to handle uncertainty, adapt to changing conditions, and make safe decisions under time pressure.

Key Research Topics

Reinforcement learning for robotic control

Transformer architectures for sequence planning

Uncertainty quantification in safety-critical systems

Efficient inference on edge devices

Computer Vision

Our perception systems understand spatial relationships, material properties, and assembly sequences in real-time. We combine multiple sensor modalities to build robust representations of the construction environment.

Key Research Topics

3D scene reconstruction and understanding

Material detection and classification

Human activity recognition for safety

Multi-sensor fusion

Edge Computing

We build distributed inference pipelines that operate reliably in harsh job site conditions with minimal latency. Our systems are designed for power efficiency while maintaining the performance needed for real-time operation.

Key Research Topics

Model compression and optimization

Distributed computing architectures

Low-latency inference systems

Robust networking for industrial environments

Robotic Control

Our motion planning and manipulation algorithms enable precise, safe interaction with construction materials. We develop controllers that can handle the variability and uncertainty inherent in real-world construction tasks.