Tanvir Ahmed Khan

Recent EEE graduate from BUET. Passionate about machine learning, NLP, and computer vision, aiming to pursue a Ph.D. in Fall 2026.

Weed infestation significantly hampers agricultural productivity in Bangladesh, especially for small and marginal farmers. AgroBot is a low-cost, autonomous precision weeding robot designed using ROS and AI-based weed detection (YOLOv5) to address this problem.

Key features:

  • Vision-guided weeding with YOLOv5 for real-time plant classification.
  • Automated navigation using wheel encoder + IMU fused by an Extended Kalman Filter (EKF).
  • Targeted herbicide spraying mechanism to minimize chemical use.
  • Cost-effective deployment using Raspberry Pi and ESP32 (~25,140 BDT total cost).

Our system achieved a mAP50 score of 0.763 with <0.2ms detection latency, and showed a 90% reduction in herbicide usage during tests.

This work contributes to:

  • Supporting small-scale farmers with affordable robotics.
  • Reducing labor and environmental impact from traditional weed control.
  • Advancing Bangladesh’s national goals in food security, smart agriculture, and technological innovation.

📄 View Poster Below: Methodlogy figure

Updated: