Mathematical Models TRIZ Cases

Segmented waypoint planning adapts to nearby people and objects in real time, cutting navigation energy use while preserving safe interaction.

Coherence checking between static and dynamic environment knowledge helps autonomous systems detect abnormal events and generate corrective objectives.

Predicting nearby vehicle intent and trajectory helps autonomous motion planning account for interaction-driven collision risks.

Multi-sensor vehicle monitoring detects approaching animals around campsites and triggers targeted lights or sounds to improve camper safety.

A decision system uses driver and autonomy capability assessment to hand off vehicle control safely during unexpected road events.

By separating road from non-road image regions, this case cuts trajectory-planning compute load while improving convergence and path stability.

Realistic sensor noise is added to 3D scene data to generate training-ready autonomous driving simulations with higher fidelity and lower latency.

Multiple navigation probability distributions and time-step search make traffic simulation more human-like while lowering agent collisions.

CNN-based path distributions speed autonomous parking path planning in unknown maps while cutting memory use and adapting to obstacles in real time.

Simulated domain images train a deep neural network to estimate magnetic parameters from observed patterns, bridging experiment and theory.

Information gain and Bayesian scenario ranking cut autonomous vehicle test cases while preserving accurate performance and safety evaluation.

Machine learning combines signal-light temporal states with vehicle orientation cues to predict driving intent despite occlusion and angle changes.

Machine learning combines vehicle, telematics, and mobile data to reconstruct collisions faster and with more consistent fault assessment.

A causal control model updates internal parameters when environment properties change, reducing compute load while keeping control settings responsive.

Neural networks predict how target agents react to a planned vehicle path, improving interactive driving accuracy while limiting compute load.

Uses vehicle and driver data with log-probability matching to evaluate steering comfort online during autonomous vehicle take-over.

Real-time traffic and sensor data guide lane choice, while delayed travel-time feedback helps the ML model reduce route time and discomfort.

Latent state inference and graph-based agent modeling improve autonomous path planning under cooperative and aggressive driving behavior.

AI estimates critical branch flows to shrink SC-OPF complexity, enabling faster and secure power grid operation under contingencies.

Combining model-based SOC estimation with adaptive current correction reduces long-term drift and keeps battery tracking within confidence limits.

Updates battery model parameters only under defined conditions to track aging, improve power prediction, and support safer storage operation.

Impedance sensing and sensor fusion let a machine learning model detect hand grip blockage and improve antenna selection, tuning, and power control.

A hybrid RL controller combines neural action proposals with reward-and-risk analysis to improve autonomous driving decisions under uncertainty.

Reinforcement learning adjusts in-vehicle climate and lighting from user reactions to cut driver distraction and manual setting changes.

Standardized scenario indicators turn context-dependent ADAS and ADS KPIs into comparable scores across real-world and simulated driving tasks.

Multi-particle state modeling with reinforcement learning adapts grid power dispatch under uncertainty and improves control with real outcomes.

During navigation, a mapping app classifies driving event sounds as real or artificial, then masks or flags false alerts to reduce driver distraction.

When ECU speed data is missing, a trained model maps GPS location data to equivalent real-time ECU speed for more accurate fleet alerts.

Frequency-filtered event camera data and active light projection help detect slow or static road objects and calculate time to contact.

Real-time sensor fusion and driver feedback help personalized ACC learn following gap preferences from cabin, weather, and traffic data.

Unexpected acceleration changes are classified by threshold and sensor context to detect minor and major vehicle collisions with fewer false alarms.

Intermediate radar updates keep object tracks current between slower lidar and image cycles, cutting perception latency for dynamic vehicles.

A disaggregation model maps splitting and merging network islands to pinpoint interruption sources at component level and improve predictive maintenance.

Reinforcement learning and neural networks turn vehicle motion data into wheel alignment signals for real-time steering geometry adjustment.

Procedural instance timing improves causal attribution of control settings to environment responses with less historical data and faster adaptation.

Data-driven trajectory models simulate human driving intentions and vicinal traffic scenarios to improve collision prediction with dynamic obstacles.

Predicting surrounding agents' future behavior lets a main agent choose safer, higher-reward actions in multi-agent environments.

Machine learning on DTCs, odometer readings, and battery voltage predicts likely vehicle failures early to cut warranty repairs and recalls.

Leakage inductance current signals and a Levy-flight sparrow search help a deep belief network avoid local optima in DC/DC converter fault diagnosis.

A maneuver filter blocks disallowed autonomous driving actions from ML probability outputs, improving safety without model retraining.

At unsignalized intersections, virtual vehicle prediction helps autonomous control handle blocked views and avoid collision risks.

Latent state inference and vehicle interaction graphs help autonomous driving predict agent behavior and plan safer paths in complex traffic.

Joint prediction and planning uses learned interaction models and cost scoring to improve autonomous vehicle trajectory accuracy with low latency.

Uses lightweight machine learning on historical equipment load data to forecast overload capability and adjust parameters before overloads occur.

Multiple capacitive sensing zones and Bayesian signal models improve windshield moisture detection while reducing false wiper activation.

Sequential smartphone sensor fusion tracks vehicle micro-activities to distinguish drivers from passengers with fewer false alarms.

ML-based portable device localization predicts user actions and actuates vehicle components for reliable keyless access and control.

Fusing object state data with geographic context helps predict multiple trajectories faster and more accurately for autonomous vehicle control.

Fused LIDAR and map data predict future occupancy by object type, helping autonomous vehicles choose safer trajectories with lower compute and storage load.

Probabilistic control functions combine driving objectives to generate human-like vehicle paths while adapting to uncertainty and passenger comfort.