Integrity Management of the Reachable Space With Lane Grid Maps

Sanchez, Corentin; Xu, Philippe; Bonnifait, Philippe; Armand, Alexandre · 2023 · Crossref

DOI: 10.1109/tiv.2022.3225071

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 critical challenge of ensuring the integrity of situation understanding for autonomous vehicles, specifically focusing on the prediction of reachable spaces for interacting road users. Incorrect predictions can lead to hazardous decision-making, such as an ego vehicle misjudging the time available to maneuver because it underestimates the speed or position of another vehicle. The authors propose a method to manage the integrity of this prediction information using Lane Grid Maps (LGMs), a spatial representation based on the topological layer of high-definition maps. The core contribution is demonstrating how the spatial sampling step of the grid can be tuned to control the integrity of both current situation understanding and future predictions, even in the presence of imperfect object detection and prediction modules. The methodology employs LGMs, which discretize lanes of interest into contiguous cells along their longitudinal direction. This representation is more compact and scalable than traditional occupancy grids, as it focuses only on areas relevant to the ego vehicle’s interactions. The paper formalizes integrity through the False Negative Rate (FNR), which measures the frequency of occupied cells incorrectly classified as free—a dangerous error that could lead to collisions. The authors show that FNR can be managed by adjusting the grid’s sampling step; larger cells provide coarser, more cautious information that reduces the risk of false negatives caused by sensor noise or localization errors. To handle interactions, the paper introduces "Lane Neutralization," a concept where physical constraints among road users (e.g., a vehicle cannot overtake another on a single lane) are used to refine reachable sets. Experiments were conducted in Compiègne, France, using three vehicles equipped with Lidar sensors and high-precision GNSS ground truth. The study evaluated the integrity of LGMs under various levels of simulated detection noise and localization uncertainty. The results demonstrate that the False Negative Rate decreases as the spatial sampling step increases, confirming that grid resolution can be used as a control parameter for integrity. Specifically, the authors found that for a given Target Integrity Risk (TIR), such as 0.3%, the required sampling step scales with the level of uncertainty in the system. For instance, with a localization error standard deviation of 0.5 meters, a sampling step of 3.0 meters was necessary to meet the integrity requirement, whereas a step below 1 meter sufficed for errors smaller than 0.2 meters. The study also validated that this sampling strategy remains effective when extending integrity management to predicted reachable spaces, ensuring that the predicted bounds of other vehicles do not introduce misleading information. The significance of this work lies in providing a practical mechanism for autonomous vehicles to guarantee the reliability of their situational awareness and predictions. By linking grid resolution to integrity metrics, the approach allows vehicles to adapt their representation of the environment based on real-time uncertainty levels, thereby preventing hazardous decisions caused by imperfect perception or prediction. This contributes to the field of autonomous driving by offering a formalized way to handle the integrity of navigation information at the tactical level, ensuring that decision-making systems are not misled by erroneous spatial data.

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