Organizations: Department of Electrical and Computer Engineering, University of the Bundeswehr Munich, Werner-Heisenberg-Weg, Neubiberg, 85579, Germany.
This paper addresses emergency-braking decisions for a redundant safety channel in automated vehicles. The channel is responsible for identifying hazards from sensor data and determining whether braking is necessary, given the constrained computational resources of embedded hardware. We present an emergency-braking approach that uses a predicted motion corridor to filter \ac{lidar} measurements and then decide whether the ego vehicle must brake. \ac{pmcm} constructs a curvature-aware boundary from the planned path and vehicle geometry so that the monitored region contains the full swept body on both straight and curved paths. \ac{gdeta} then decides whether the ego vehicle must command an emergency brake for objects inside this boundary, scoring geometric occupancy and temporal urgency independently. Hard gates permit braking when either source is conclusive; soft-OR arbitration combines partial evidence when neither gate is active. Consecutive-scan confirmation reduces sensitivity to transient measurement noise. Corridor coverage is evaluated in controlled collision simulations and recorded campus driving. Braking decisions are evaluated in Monte Carlo simulations of lead-vehicle braking and transverse-crossing encounters.
Figures & tables
Figure 1: Automated vehicle technical stack. The fail-safe stack is NeFiMMORE with a dedicated LiDAR and brake authority. The main stack is the HLAS with the full sensor suite, steering, and powertrain. The only runtime coupling is the planned trajectory toward the monitor and a warning plus brake request back to the vehicle Pan et al. (2026) .
Figure 2: Experimental vehicle sensor setup. Our NeFiMMORE system uses a low-mounted front 3D LiDAR to perceive the near-field environment.
Figure 3: Predicted motion corridor (marked in red) in representative near-field scenes. Longitudinal depth follows the current speed; lateral bounds follow the planned body sweep Pan et al. (2026) .
Figure 4: Overview of the runtime processing pipeline implemented in NeFiMMORE .
Figure 5: Predicted reference path γm(s) in a curvilinear frame, together with the motion-corridor envelope obtained from the vehicle-corner tracks Pan et al. (2026) .
Figure 6: Construction of the visible monitoring boundary from a predicted motion corridor: (a) boundary preconstruction; (b) visible reduction; (c) occlusion filtering Pan et al. (2026) .
Figure 7: Local azimuth window of two successive visible vertices. Point pj lies inside the filtering triangle; pj′ lies outside the admissible range Pan et al. (2026) .
Measure
Parallelogram
Two-point corridor
PMCM
FNR
0%
57.74%
0%
TPR
100%
42.26%
100%
CR
36.1%
1.03%
0%
Table 1: Comparison of FNR , TPR , and residual CR for three monitoring masks on the same controlled collision series. Adapted from Pan et al. (2026) .
Figure 8: Bird’s-eye view of a simulated corridor-edge collision case. The parallelogram mask, the PMCM corridor, and the first-detection location pos. A are indicated Pan et al. (2026) .
Figure 9: Online PMCM corridor on recorded campus driving, shown in a camera view and a bird’s-eye overlay Pan et al. (2026) .
Property
Established lead braking
Transverse crossing
Trials / seed
200,000 / 10
200,000 / 10
Horizon
60s
12s
Ego speed
3 – 30km/h
3 – 30km/h
Target body
3.5×2.0m vehicle
0.5×0.4m target
Target motion
0 – 30km/h ; 3 – 7.5m/s2 braking
vs=0 ; ∣vd∣=0.5 – 3m/s
Encounter timing
gap 0.5 – 15m ; brake onset 0 – 5s
centerline-arrival offset −3 – 3s
Table 2: Object-level Monte Carlo scene design. Continuous quantities are sampled independently from uniform distributions.
Computational models of careful and competent human drivers are essential for scenario-based evaluation of automated driving systems (ADS). However, most existing safety reference models primarily focus on longitudinal braking, neglecting the role of evasive steering in human collision avoidance. This paper proposes a hybrid Fuzzy-Safety Model (FSM-H) that integrates longitudinal mitigation and lateral avoidance within a unified behavioral framework. The braking component is governed by Proactive Fuzzy Safety (PFS) metrics, representing the erosion of longitudinal safety margins, while the steering component is driven by Criticality Fuzzy Safety for lane-change (CFS-LC), capturing lateral conflict severity and maneuver feasibility. A finite-state architecture models the sequential escalation from nominal driving to braking and, when necessary, to evasive steering, incorporating perception-reaction time and lane-check delays to reflect human decision processes. The model is evaluated in reconstructed high-criticality cut-in scenarios and compared with braking-only and steering-only reference strategies. Results show that the hybrid approach expands the preventability envelope while maintaining behavioral plausibility and computational tractability. The proposed framework provides a transparent and explainable human reference model suitable for simulation-based ADS safety benchmarking and regulatory assessment.
Riccardo Donà, Konstantinos Mattas, Biagio Ciuffo
Joint Research Centre for the European Commission, Ispra, IT
Emergency collision avoidance under extreme driving conditions demands safety-critical control that accounts for both obstacle proximity and vehicle dynamic stability over a future time horizon, yet existing methods often rely on instantaneous or local safety evaluations. This paper proposes a safe reinforcement learning framework guided by a Hamilton-Jacobi (HJ) reachability based motion safety set that provides forward-looking safety supervision for constrained policy optimization. Specifically, a unified signed safety function is formulated by combining geometric collision margins and chassis stability limits, and is then extended through reachability analysis into a finite-horizon motion safety set that characterizes whether safety can be maintained under future vehicle state evolution. To enable practical computation, the motion safety set is approximated from offline extreme driving data, mitigating the computational burden of grid-based HJ solvers. The learned motion safety set is then embedded as a continuous safety cost into a constrained Markov decision process, and a PID-Lagrangian policy optimization scheme is employed to adaptively regulate the Lagrange multiplier for safety constraint enforcement. Simulation and real-vehicle experiments on low-adhesion obstacle-avoidance scenarios demonstrate that the proposed method achieves higher goal-reaching rates, produces smoother avoidance maneuvers, and maintains larger unified safety margins than baseline methods.
Yuhong Jiang, Shiyue Zhao, Junzhi Zhang +4
School of Vehicle and Mobility, Tsinghua University, Beijing, China · State Key Laboratory of Intelligent Green Vehicle and Mobility, Tsinghua University, Beijing, China
Intent misinterpretation during vehicle interactions causes recurring planning failures. We study a decision layer in which a language-guided intent module reads structured descriptors, computes a smoothed intent-geometry divergence score, and gates the planned maneuver before commitment, upstream of a corridor envelope. On a replayed off-road departure and four crash clips under a frozen, disclosed implementation, gating is the only layer that repairs the plan: on the main case it fires 72 ms after the drift onset but 161 ms before the corridor exit, keeping the trajectory in the corridor in all ten replays. The first calibration draws nine false triggers in 5.9 minutes, each from scoring uncertainty as half a conflict; a preregistered redesign treating uncertainty as abstention cuts this to 0.341 per minute. Two ablations bound the model's contribution: the full score detects fastest on four of five failures under the deployed eligibility, three of five against the unvetoed rule (000871 by one cycle; 000228 by a pre-onset fire on an uncertain stretch that five clips cannot classify as signal or coincidence; dropping the confidence term costs two detections), while on in-domain tracks at equal false positives the geometric rule more than triples its detection. The evidence supports the gating mechanism; the model's demonstrated roles are the fastest detection on these failures and an uncertainty veto on the geometric rule.
Cong Xu, Ravi Sankar
iCONS Lab, Department of Electrical Engineering, University of South Florida, Tampa, FL 33620, USA