Invention:
This invention consists of an algorithm that utilizes computer vision and thermal imaging cameras to detect, track, and alert for hazardous fall events in environments with exposed and potentially unstable rock walls or structures. The Automated Rockfall Recognition system utilizes mathematical, non-machine-learning methods to identify rockfall events by using thermal video system with low resolution and low framerate thermal video inputs. It can alert users to rockfalls in environments with low visibility and other conditions that may hinder other monitoring technologies (e.g., fog, dust, darkness), helping improve mine safety.
Background:
Mine sites are active at all times of day and often have visibility obscured by dust, precipitation, or other weather conditions. Rockfall is a common hazard in mining and can include large and easily visible singular rocks, slides of aggregate material, or falling objects. Open pit mines and similar environments that have irregular structures of slope surfaces can make rockfall drastically change direction. Falling objects can deflect or dislodge other materials that may cause further damage. There is a need for better mining safety solutions that can predict rockfall occurrences.
Applications:
- Mining safety
- Transportation
- Rockfall detection
Advantages:
- Uses a non-machine-learning approach
- Enhanced detection in low-visibility conditions
- Potential to provide time-of-failure predictions
- Does not require retraining data
