Top 5 Finish at eYRC Robotics Competition
Achieved 5th place in the national-level e-Yantra Robotics Competition 2023-24 with a project involving machine learning, image processing, and GIS.
Top 5 Finish at eYRC Robotics Competition
Competition Overview
The e-Yantra Robotics Competition (eYRC) is a prestigious national-level competition organized by IIT Bombay that challenges students to develop innovative robotics solutions. Our team achieved a remarkable 5th place finish in the 2023-24 edition under the GeoGuide theme.
Project: Team Vanguard
Our project, aptly named Vanguard, represented the cutting edge of autonomous robotics, combining multiple disciplines including machine learning, computer vision, and geospatial technologies.
Project Objectives
The challenge required developing an autonomous robot capable of:
- Autonomous Navigation: Following predefined paths with precision
- Object Detection: Identifying and classifying objects in real-time
- Spatial Mapping: Real-time position tracking and mapping using GIS
Technical Implementation
Autonomous Navigation System
- Line Following: Implemented using computer vision and WebSocket communication
- Path Planning: Integrated Dijkstra's Algorithm for optimal route calculation
- Real-time Control: Responsive movement system with obstacle avoidance
Machine Learning Pipeline
- Transfer Learning: Fine-tuned pre-trained models for object classification
- Optimization: Achieved high accuracy on 100x100 low-resolution input images
- Real-time Inference: Optimized for on-device processing with minimal latency
GIS Integration
- QGIS Mapping: Real-time robot position plotting and visualization
- Spatial Analysis: Geographic data processing for route optimization
- Data Visualization: Interactive maps showing robot trajectory and detected objects
Key Technical Achievements
Algorithm Development
- Successfully implemented Dijkstra's pathfinding algorithm for optimal route planning
- Developed efficient graph traversal methods for dynamic environments
- Created adaptive algorithms that respond to real-time sensor data
Computer Vision Excellence
- Fine-tuned Convolutional Neural Networks (CNNs) for object detection
- Achieved high classification accuracy despite resource constraints
- Optimized models for embedded systems deployment
Real-time Systems
- Built robust WebSocket communication for robot control
- Implemented real-time data streaming for position tracking
- Created responsive control systems with minimal latency
Learning Outcomes
This competition provided invaluable experience in:
Interdisciplinary Integration
- Robotics Engineering: Hardware control and sensor integration
- Machine Learning: Model training, optimization, and deployment
- Geospatial Technology: GIS analysis and mapping systems
- Software Engineering: Real-time systems and communication protocols
Problem-Solving Skills
- Complex system design and architecture
- Real-time debugging and optimization
- Team collaboration and project management
- Performance optimization under constraints
Competition Impact
The eYRC experience significantly strengthened my foundations in:
- Robotics and Automation
- Machine Learning Applications
- Spatial Data Processing
- Real-time System Design
This competition served as a launching pad for my later research work at e-Yantra, IIT Bombay, where I continued to explore the intersection of AI and robotics.
Technologies and Tools
- Programming: Python, C++, JavaScript
- Machine Learning: TensorFlow, Keras, OpenCV
- GIS: QGIS, Geospatial libraries
- Communication: WebSocket, Real-time protocols
- Hardware: Embedded systems, Sensors, Actuators
Team Achievement
Our 5th place finish out of hundreds of participating teams nationwide represents:
- Excellence in technical implementation
- Strong teamwork and collaboration
- Innovative problem-solving approach
- Successful project delivery under pressure
This achievement marked a significant milestone in my robotics journey and laid the foundation for my continued work in AI and automation technologies.