Problem Statements

In coal mining, determining the quality of coal (like ash content, moisture, calorific value, etc.) usually takes time because it depends on lab testing. This delay can slow down decision-making and affect blending, pricing, and dispatch.

Build an AI-based software system that can predict coal quality instantly using historical lab data, mine location, geological information, and real-time sensor inputs (if available). Instead of waiting for lab results every time, the system should estimate coal composition quickly and accurately, helping teams make faster decisions.

• Faster coal quality assessment without waiting for lab reports
• More consistent and optimised coal blending
• Better pricing decisions based on predicted quality
• Reduced dependency on manual and time-consuming testing
• Improved customer satisfaction due to consistent coal supply quality
• Optional Enhancement: Predicting ash %, moisture %, volatile matter, fixed carbon, GCV (Gross Calorific Value)
• Integration with mine dispatch systems
• Continuous learning model that improves as more data is added

In underground mines, there are certain areas that become too dangerous or impossible for workers to enter due to reasons like toxic gas, poor visibility, roof collapse risk, waterlogging, fire, or lack of ventilation. In such situations, sending people for inspection puts lives at serious risk.

Develop an Unmanned Ground Vehicle (UGV) capable of operating within underground mines and monitoring areas where human entry is unsafe or not possible. The UGV should be able to travel through narrow and difficult mine passages while collecting important information such as gas levels, temperature, humidity, structural condition, water presence, and live video. The system should help mine operators remotely inspect hazardous zones, assess risks, and make quick decisions without exposing workers to danger.

• Safe monitoring of underground mine zones where workers cannot reach
• Reduction in risk to human life during inspection and emergencies
• Real-time collection of critical data such as gas concentration, temperature, and visual conditions
• Faster identification of hazards like gas leakage, roof fall risk, flooding, or fire-prone conditions
• Better emergency response and rescue planning
• Improved overall underground mine safety and operational awareness

Modern mining operations are complex, dynamic, and high-risk, making it difficult to test new strategies directly in the field. Imagine having a virtual replica of a mine — a "digital twin" — where engineers and planners can experiment with different scenarios, optimise operations, and predict outcomes without disrupting real-world activities or compromising safety.

Build a digital model that mirrors actual mining operations and helps simulate different scenarios, enabling teams to make data-driven decisions, test interventions safely, and plan operations with greater precision.

• Better planning and fewer costly mistakes
• Ability to test "what if" scenarios safely
• Improved productivity and cost savings
• Smarter long-term decisions

All mines operate under strict safety regulations. However, continuous manual monitoring to ensure compliance — such as wearing personal protective equipment (PPE) and following safe practices — is challenging, error-prone, and resource-intensive.

Design and develop an AI-powered system that leverages camera feeds and computer vision techniques to monitor and detect safety compliance in real time automatically. The system should be capable of identifying whether workers are wearing required safety gear (e.g., helmets, gloves, vests) and adhering to defined safety protocols.

• Fewer workplace accidents
• Automatic monitoring instead of manual checks
• Quick alerts for unsafe situations
• Stronger safety culture