Research

NavSense Lab conducts independent research at the intersection of edge computing, AI, and autonomous systems. Current research areas:

Edge AI & TinyML

Deploying deep learning on microcontrollers and edge devices for real-time inference in constrained IoT environments.

• On-device inference with ONNX and TensorFlow Lite
• Model compression, pruning, and quantization
• KITTI-based visual localization on edge hardware

IoT Localization

Fingerprinting-based and ray-tracing-assisted localization for indoor and vehicular environments, including crowdsourced approaches.

• RF fingerprinting and channel modeling
• Ray-tracing-assisted indoor positioning
• Crowdsourced-based vehicular localization

Federated & Continual Learning

Privacy-preserving distributed training with path-coordinated continual learning for non-stationary environments.

• Path-coordinated continual learning for IoT streams
• Federated learning with differential privacy
• Non-IID data handling in distributed edge settings

Autonomous & Multi-Robot Systems

Multi-robot task allocation, swarm coordination, and path planning for autonomous edge deployments.

• Decentralized task allocation protocols
• Swarm intelligence for edge AI coordination
• LLM-orchestrated multi-agent systems (Qwen, DeepSeek)