Adaptive cruise control adjusts settings for different driving scenarios
Cross-Scenario Controller Parameter Transformation from Real-World Driving Data through Flow Matching
Machine Learning
Summary
Many driver assistance systems feel strange because they don’t match how individual people drive. This paper shows how to change settings for cruise control from one driving situation to another based on only a few examples of how a person drives in the first situation. The authors use a math technique called flow matching to predict how a driver would behave in a new scenario. Their method is smarter than simple ways that just shift settings linearly and includes ways to capture each driver’s unique habits.
What this means in practice
- •For automotive engineers: Create adaptive cruise control systems that automatically adjust to different driving situations based on limited driver data for improved user acceptance.$Commercial implications: Enables development of personalized driver assistance products that feel more natural and encourage use by tailoring control parameters to individuals.
- •For vehicle software developers: Improve driver assistance software by integrating models that translate driver behavior patterns between various road and traffic scenarios.
Authors
Christian Zink, Maik Pfefferkorn, Eric Lenz, Felix Kallmeyer, Rolf Findeisen
Abstract
Advanced Driver Assistance Systems have become increasingly widespread, aiming to support drivers, reduce workload, and improve traffic safety. However, many systems are rarely used because their behavior feels unnatural and does not match individual driving behavior. The growing availability of large-scale, real-world human driving data enables more human-like and driver-individual assistance functions. This paper formulates the adaptation of Adaptive Cruise Control parameters as a driver-individual scenario-to-scenario transport problem, transforming controller parameters from one driving context to another. Using flow matching, we train a generative model to predict the target-scenario behavior of previously unseen drivers from only a few demonstrations in a source scenario. We compare the approach with a linear parameter-shift baseline and evaluate flow-matching variants incorporating driver-consistent coupling, driver embeddings, and classifier-free guidance based on driving profiles extracted from trajectory data using a variational autoencoder.