Combining Stochastic and Scenario Model Predictive Control to Handle Target Vehicle Uncertainty in an Autonomous Driving Highway Scenario

Brüdigam, Tim; Olbrich, Michael; Leibold, Marion; Wollherr, Dirk · 2018 · OpenAlex-citations

DOI: 10.1109/itsc.2018.8569909

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of safe and efficient trajectory planning for autonomous vehicles in uncertain highway environments. The authors identify two distinct types of uncertainty regarding surrounding "target vehicles" (TVs): the possibility of multiple future maneuvers (e.g., lane keeping vs. lane changing) and the execution uncertainty within any specific predicted maneuver. Relying on only one type of uncertainty leads to either neglected risks or overly conservative motion planning. To resolve this, the authors propose a combined Stochastic and Scenario Model Predictive Control (S+SC MPC) method that balances safety and performance through adjustable risk parameters. The methodology integrates Scenario MPC (SCMPC) to handle discrete maneuver possibilities and Stochastic MPC (SMPC) to manage continuous trajectory deviations. The system uses linear, discrete point-mass models for both the ego vehicle and target vehicles. Safety is enforced via an elliptical constraint around the TV. To account for multiple maneuvers, the method draws samples based on maneuver probabilities to create a "virtual" combined ellipse that covers all potential future states. This reduces the number of samples required compared to pure SCMPC. Simultaneously, SMPC employs chance-constraints to ensure that the probability of violating safety constraints due to execution errors remains below a specified threshold. The probabilistic constraints are reformulated into deterministic inequalities using prediction error covariance matrices, allowing for efficient optimization. The study demonstrates the effectiveness of the S+SC MPC approach through a simulation of a two-lane highway scenario. The results show that the combined method accurately predicts target vehicle behavior while limiting the conservativeness of the ego vehicle’s trajectory. By separating maneuver uncertainty (handled by SCMPC sampling) from execution uncertainty (handled by SMPC chance-constraints), the approach avoids the high computational cost of excessive sampling and the imprecision of ignoring maneuver variability. The adjustable risk parameters allow for a tunable trade-off between safety assurance and driving efficiency. The significance of this work lies in providing a robust framework for autonomous driving control that explicitly handles complex, multi-layered uncertainties. The S+SC MPC method offers a practical solution for real-time trajectory planning, ensuring that safety constraints are satisfied with a guaranteed probability while maintaining efficient vehicle operation. This contributes to the field by advancing the state-of-the-art in MPC applications for autonomous systems, particularly in scenarios involving dynamic interactions with other traffic participants.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• situational awareness

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Theoretical Contribution: computational model