AI summary by TwinMind
Why this "falling rocks" sign is more important than most
Tom Scott
5m 3s
Location and Problem
- Brienz, Switzerland - small alpine village surrounded by mountains
- Hillside slowly sliding/slipping down, unstable terrain
- Large rocks tumble down at 10-20 km/h sideways across road
- Village safe from rockfalls due to distance, but road vulnerable
- Massive protection barriers too expensive along entire route
- Low but non-zero probability of boulder hitting car
Radar Detection Solution
- Automatic rockfall detection system with radar and traffic lights
- Radar detects rockfall in defined area, triggers alarm
- Traffic lights turn red when rockfall detected
- Radar tracks rockfall trajectory and landing location
- System responses:
- Rockfall stops/ends in dam: lights return to normal, road safe
- Rockfall near road: lights switch to flashing orange ("drive carefully")
- Boulder may be on road or road damaged
Technical Specifications
- Radar advantages: works in all weather/visibility conditions
- Penetrates fog, rainfall, snowfall - operates day/night
- Traffic lights positioned 450 meters apart
- 450m = dangerous area for traffic
- 40 seconds from rock detachment to reaching road
- Just enough time for cars to escape at certain velocity
- Cyclists prohibited due to insufficient escape speed
System Calibration Challenges
- Cannot trigger on all movement detection
- Birds and drones would cause false alarms
- Calibration critical: balance false positives vs false negatives
- False positives = road closed half the time
- False negatives = cars hit by boulders
- Rain fronts can appear similar to rockfalls on radar
Implementation and Performance
- Radar installed 2017
- 1 year test period before going live with traffic lights
- 5,500 events detected since installation
- Very rare false alarms
- Active site perfect for algorithm development/tuning
- Engineering: radar/stoplight setup easy part
- Complicated part: calibrating detection algorithms
