Predictive Force-Centric Emergency Collision Avoidance

Fors, Victor; Anistratov, Pavel; Olofsson, Björn; Nielsen, Lars · 2021 · Crossref

DOI: 10.1115/1.4050403

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 presents a predictive, force-centric controller designed for emergency collision avoidance in autonomous vehicles. The research addresses the challenge of executing critical maneuvers that simultaneously maximize braking to reduce collision severity and steer to avoid obstacles, all while maintaining vehicle stability at the limit of tire-road friction. The motivation stems from the need for robust autonomous safety systems that can handle complex scenarios, such as avoiding moving obstacles like pedestrians or animals, without relying on human driver intervention. The proposed control architecture consists of two primary stages: a high-level motion planner and a low-level acceleration follower. The motion planner utilizes a receding-horizon optimization approach based on a friction-limited particle model. It transforms the vehicle dynamics into road-aligned coordinates to efficiently compute acceleration-vector references (longitudinal and lateral) that minimize velocity over a planning horizon while respecting road boundaries and obstacle constraints. The optimization problem is solved numerically using CasADi and IPOPT, with specific penalties for boundary violations and friction slack variables to ensure feasibility. The low-level controller then translates these acceleration references into specific actuator commands—steering wheel angle and individual wheel braking torques—using a modified Hamiltonian algorithm. This algorithm maximizes the available tire forces to match the desired global acceleration and yaw moment, effectively distributing forces across the four wheels. The controller was evaluated in challenging scenarios involving stationary and moving obstacles crossing the road. Results demonstrate that the system successfully balances braking and avoidance maneuvers. Specifically, when encountering a moving obstacle, the controller exploits widening gaps to apply more aggressive braking, thereby reducing impact speed. The vehicle exhibited well-behaved dynamics regarding steering angles, body slip, and path tracking. Furthermore, the study confirms the controller’s real-time capabilities, showing that the computational overhead of the receding-horizon planning and force distribution is manageable for online implementation. The significance of this work lies in its integration of optimization-based planning with force-centric control to achieve optimal performance at the friction limit. By prioritizing braking within the constraints of avoidance, the system enhances safety by reducing potential crash severity. The methodology provides a viable framework for autonomous emergency maneuvers, offering a robust solution that adapts to evolving situations through continuous replanning and precise force vectoring.

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.

• perception action locomotion