Research and publications

A collection of my published research in robotics, embodied AI, and intelligent systems.

Title page of VisualNav: Visually Grounded Natural Language Crawler Robot Navigation
4th International Conference on Image Processing and Robotics (ICIPRoB 2026)2026

VisualNav: Visually Grounded Natural Language Crawler Robot Navigation

Authors: Lakmina P. Gamage, Haritha Weerathunga, Vishwani Geeganage, Chinthaka Premachandra, B. H. Sudantha

See more

Abstract

This paper compares two vision-language-action architectures for natural-language crawler robot navigation. It evaluates an end-to-end vision-language model against a modular system that separates intent extraction, open-vocabulary visual grounding, and control. The modular approach demonstrates stronger zero-shot generalization and lower inference latency, offering a practical route for embodied AI on resource-constrained edge devices.

  • Robot Navigation
  • Robotics
  • Vision Language Models
  • Vision-Language-Action Models
  • Visual Grounding
Read publication
Title page of Reinforcement Learning-Based Autonomous UAV Navigation in Indoor Environments Towards Human Detection
4th International Conference on Image Processing and Robotics (ICIPRoB 2026)2026

Reinforcement Learning-Based Autonomous UAV Navigation in Indoor Environments Towards Human Detection

Authors: Vishwani Geeganage, Lakmina Gamage, Haritha Weerathunga, Chinthaka Premachandra, B. H. Sudantha

See more

Abstract

This paper presents a low-cost autonomous UAV navigation framework for indoor search-and-rescue missions in GPS-denied environments. Using monocular vision, YOLOv8n human detection, and deep reinforcement learning, it compares DQN and PPO navigation strategies. The PPO agent achieves more reliable navigation and is validated through sim-to-real transfer on a DJI Tello drone without expensive sensing hardware.

  • Proximal Policy Optimization
  • Reinforcement Learning
  • Sim-to-real Transfer
  • Unmanned Aerial Vehicles
  • YOLOv8n
Read publication