Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Priority-Driven Safe Model Predictive Control Approach to Autonomous Driving Applications

Francesco Prignoli, Ying Shuai Quan, Mohammad Jeddi, Jonas Sjöberg, Paolo Falcone

TL;DR

The paper tackles the challenge of maintaining safety in autonomous driving when disturbances cause unknown or changing constraints to become infeasible. It introduces a priority-driven softened Safe MPC framework that selectively relaxes lower-priority comfort constraints while preserving hard safety constraints, using a learning-based NN to approximate the necessary constraint relaxations. The method supports multiple relaxation modes and employs NN-based feasibility indicators to choose which constraints to relax in real time, demonstrated on ACC and automated lane-change scenarios with real-world data. Results show that the approach preserves collision avoidance and lane-keeping, achieves the desired driving behavior, and reduces run-time overhead compared to solving the full slack-augmented problem, enabling practical real-time deployment.

Abstract

This paper demonstrates the applicability of the safe model predictive control (SMPC) framework to autonomous driving scenarios, focusing on the design of adaptive cruise control (ACC) and automated lane-change systems. Building on the SMPC approach with priority-driven constraint softening -- which ensures the satisfaction of \emph{hard} constraints under external disturbances by selectively softening a predefined subset of adjustable constraints -- we show how the algorithm dynamically relaxes lower-priority, comfort-related constraints in response to unexpected disturbances while preserving critical safety requirements such as collision avoidance and lane-keeping. A learning-based algorithm approximating the time consuming SMPC is introduced to enable real-time execution. Simulations in real-world driving scenarios subject to unpredicted disturbances confirm that this prioritized softening mechanism consistently upholds stringent safety constraints, underscoring the effectiveness of the proposed method.

Priority-Driven Safe Model Predictive Control Approach to Autonomous Driving Applications

TL;DR

The paper tackles the challenge of maintaining safety in autonomous driving when disturbances cause unknown or changing constraints to become infeasible. It introduces a priority-driven softened Safe MPC framework that selectively relaxes lower-priority comfort constraints while preserving hard safety constraints, using a learning-based NN to approximate the necessary constraint relaxations. The method supports multiple relaxation modes and employs NN-based feasibility indicators to choose which constraints to relax in real time, demonstrated on ACC and automated lane-change scenarios with real-world data. Results show that the approach preserves collision avoidance and lane-keeping, achieves the desired driving behavior, and reduces run-time overhead compared to solving the full slack-augmented problem, enabling practical real-time deployment.

Abstract

This paper demonstrates the applicability of the safe model predictive control (SMPC) framework to autonomous driving scenarios, focusing on the design of adaptive cruise control (ACC) and automated lane-change systems. Building on the SMPC approach with priority-driven constraint softening -- which ensures the satisfaction of \emph{hard} constraints under external disturbances by selectively softening a predefined subset of adjustable constraints -- we show how the algorithm dynamically relaxes lower-priority, comfort-related constraints in response to unexpected disturbances while preserving critical safety requirements such as collision avoidance and lane-keeping. A learning-based algorithm approximating the time consuming SMPC is introduced to enable real-time execution. Simulations in real-world driving scenarios subject to unpredicted disturbances confirm that this prioritized softening mechanism consistently upholds stringent safety constraints, underscoring the effectiveness of the proposed method.
Paper Structure (14 sections, 2 theorems, 29 equations, 5 figures, 1 algorithm)

This paper contains 14 sections, 2 theorems, 29 equations, 5 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Notation
  3. Background
  4. Priority-driven softened Safe MPC
  5. Neural network (NN) approximation for constraint softening
  6. Relaxed consistency condition and feasibility
  7. Priority-driven soft-constrained MPC
  8. Problem formulation
  9. Vehicle Model and System Constraints
  10. Cost and system constraints
  11. Simulation Results
  12. Scenario 1: Adaptive Cruise Control with Constraint Relaxation
  13. Scenario 2: Emergency Lane Change Maneuver
  14. Conclusion

Key Result

Proposition 1

batkovic2022safe Suppose that Assumptions a:cont, a:rec_ref, a:terminal, a:unknown_constraints, and a:safe hold, and that Problem (eq:nmpc) is feasible for the initial state ${\mathbf{x}}_k$. Then, system (eq:sys) in a closed loop with the solution of (eq:nmpc) applied in receding horizon is safe (r

Figures (5)

  • Figure 1: Time sequence in Scenario 1: RU (black) lane crossing forces the ego vehicle (red) to exceed comfort acceleration limits.
  • Figure 2: Closed-loop trajectories for Scenario 1: speed (top), distance to the RU (center), and acceleration (bottom).
  • Figure 3: Time sequence in Scenario 2: RU (black) sudden cut-in results in an emergency lane change by the ego vehicle (red).
  • Figure 4: Closed-loop trajectories for Scenario 2: lateral position error (top), lateral acceleration (center), and lateral jerk (bottom).
  • Figure 5: Computation times comparison in Scenario 1 (top) and 2 (bottom) : Problem \ref{['eq:slack_opt']} vs. NN.

Theorems & Definitions (8)

  • Definition 1: Safety
  • Proposition 1
  • proof
  • Definition 2
  • Definition 3: Safety
  • Theorem 1
  • proof
  • Remark 1