NESTNext-generation Embedded Systems Research LabJoin the lab ↗
Menu
Research

Systems research, grounded in reality

We design software methods around the physical constraints that define embedded and cyber-physical systems.

Embedded AI at the Edge

Hardware-aware training, partitioning, and inference methods that bring capable machine learning to constrained edge platforms.

  • On-device intelligence
  • Distributed DNNs
  • Resource-aware inference
Edge AI processor coordinating sensors and on-device machine learning

Secure & Safe Embedded Systems

Testing, calibration, and runtime assurance for systems where timing, safety, and reliability are non-negotiable.

  • HIL / SIL testing
  • CAN anomaly detection
  • Mixed-criticality systems
Embedded controller connected to hardware-in-the-loop safety testing equipment

Autonomous Systems

Perception, planning, health monitoring, and resilient software pipelines for UAVs, vehicles, and robotic platforms.

  • UAV autonomy
  • Fault detection
  • Cyber-physical testbeds
Autonomous aerial vehicle operating in a real-world environment

Smart Energy Systems

Embedded optimization and intelligent control for grid resilience, battery energy storage, and IoT monitoring.

  • BESS integration
  • Grid analytics
  • Digital twins
Connected smart-grid and battery energy storage infrastructure

Software Reuse & DevOps

AI-assisted feature discovery and certification-aware delivery pipelines for long-lived embedded software.

  • Feature modeling
  • ESOps
  • LLMs for software engineering
Software architecture assembled from reusable embedded feature modules

Cyber-Physical Systems & Digital Twins

Model-driven integration of sensing, computation, networking, and physical processes for observable and resilient intelligent systems.

  • Digital twins
  • System co-simulation
  • Runtime monitoring
Physical embedded testbed synchronized with a real-time digital twin

Parallel Computing & Task Scheduling

Resource-aware scheduling, workload partitioning, and parallel execution across multicore, GPU, edge, and cloud platforms.

  • GPU acceleration
  • Real-time scheduling
  • Edge-cloud orchestration
Close-up of a GPU accelerator used for parallel computing workloads
How we work

Model. Build. Measure. Improve

Model

Formalize constraints, system behavior, and measurable research questions.

Build

Create software tools, simulations, and physical prototypes.

Measure

Evaluate on realistic workloads and hardware testbeds.

Translate

Share reproducible results with academic and industry partners.