Ensemble Learning TRIZ Cases
- Hybrid Sensor Fusion Architecture for Safety-Critical Autonomy
Combining rule-based and learned fusion paths improves autonomous sensing accuracy while limiting common cause failures for higher safety integrity.
- Ultrasonic Driver Detection for Multi-Device Vehicle Occupancy
Ultrasonic sensing maps mobile devices to seat positions, helping identify the driver accurately when multiple devices are present.
- In-Vehicle Function Recommendation Using Context-Aware AI
Context-aware AI uses vehicle sensor data to recommend relevant ADAS functions, improving driver awareness, safety, and usability.
- Intersection Collision Prediction Using Telematics and Map Data
Telematics and map data are combined to rate individual intersections and predict collision probability for safer route planning.
- Ammonia Fuel Cell Control for Stable Cracker Hydrogen Supply
Predictive control uses driving data to adjust ammonia supply and cracker conditions, stabilizing hydrogen output and reducing residual ammonia.
- Battery Voltage Drop Clustering for Defect Screening
Outlier filtering, battery clustering, and temperature-corrected ΔOCV dispersion improve low-voltage defect screening reliability.
- Neural Network Ensembles for Out-of-Distribution Scenario Annotation
Ensemble discrepancy checks flag out-of-distribution driving scenarios and generate annotations to improve neural network robustness.
- Radar Facial Recognition Using Golay Sequences and RF Channel Responses
RF radar with Golay-sequence channel responses and a random forest model enables facial recognition with lower power use and no lighting dependence.
- Commercial EV Energy Prediction Using Driving Position Data
Machine learning uses discharge duration and driving position data to predict commercial EV energy use more accurately under real driving conditions.
- Autonomous Vehicle Ethical Decision Training Using GAN Location Intelligence
GAN-generated moral-island scenarios train ETHNET to help autonomous vehicles make ethical driving decisions in rare complex situations.
- Smart Circuit Breaker Load Identification From Aggregate Power Signals
Combining breaker-level and mains power signals helps identify connected devices and state changes without deploying many smart plugs.
- Deposition Condition Selection Using Confidence-Bounded Regression
Balances local accuracy and global prediction reliability in deposition by switching between linear and nonlinear models using confidence intervals.
- Near-Curb Behavior Prediction for Multi-Agent Autonomous Driving
A trained model predicts near-curb behavior for vehicles, cyclists, and pedestrians across future time points to improve path planning and safety.
- Vehicle Crash Impact Detection Using Multi-Phone Verification
Multiple mobile devices align time, location, and impact data to verify vehicle crash points and reduce false alarms.