Source-linked AI summary
Beyond the Proving Ground: Independent Public-Road Testing of Assisted Lane Change Systems using LiDAR
Marcello Cellina, Akos Kriston, Antonio Migneco, Davide Maggi, Stefano Favelli, Fabrizio Re, Fabrizio Minarini, Andrea Nuovo, Riccardo Dona, Biagio Ciuffo
TL;DR
Proving-ground scenarios cover limited conditions, and geo-fencing can impede independent assessment of ALC systems. This paper uses LiDAR for public-road testing against UNECE R79 safety distances, finding six overshoots among 27 experiments, three with 99% confidence intervals.
Problem
Proving-ground tests cover limited scenarios, while geo-fenced ALC systems can make independent public-road assessment difficult.
Source: INTRODUCTION · p. 3Method
The study uses LiDAR-based measurements to independently test commercial ALC systems on public roads against UNECE R79 critical-distance thresholds.
Source: CONCLUSIONS · p. 10Results
6 of 27 ALC experiments were completed with an R79 Critical Distance Overshot; 3 of those 6 had a 99% Confidence Interval measurement of the overshot.
Source: DISCUSSION · p. 9, CONCLUSIONS · p. 10Takeaways & Limitations
Public-road LiDAR testing is precise enough to detect critical system behaviour that may be insufficiently represented in conventional type-approval testing and could support independent market surveillance.
Source: CONCLUSIONS · p. 10The evaluation tested a single vehicle and was limited by measurement and temporal annotation precision.
Source: CONCLUSIONS · p. 10
Abstract
from arXiv · showhide
Testing of commercial Advanced Driver Assistance Systems is essential to ensure safety and compliance during type approval and in service operation. However, proving ground scenarios may not reflect real world driving complexity, while geo fencing can require manufacturer collaboration and limit assessment independence. This work presents a methodology for independently testing Assisted Lane Change systems on public roads. A campaign on the A31 French motorway used a test vehicle equipped with a LiDAR based vehicle detection and tracking system. Tests covered combinations of inter vehicle distance and speed between the test vehicle and the take over vehicle. Real time kinematic global navigation satellite system receivers assessed detection and tracking performance. Recorded lane change trajectories were compared with the lane change suppression requirements of UNECE Regulation Number 79. Of 27 predefined lane change manoeuvres, 18 were completed and 9 suppressed. In 6 cases, the system allowed manoeuvres that did not meet regulatory minimum distance requirements. In 3 cases, the deviation remained statistically significant after accounting for measurement uncertainty. To the authors knowledge, this is the first public road campaign designed to assess Assisted Lane Change compliance with Regulation Number 79 safety distance requirements. The results demonstrate the suitability of LiDAR based sensing for this purpose. The methodology can support market surveillance and future regulatory revisions by revealing real world behaviours not covered by approval procedures.
INTRODUCTION
The paper presents a LiDAR-based methodology for independently testing Assisted Lane Change systems on public roads, addressing limits of proving-ground assessment and barriers to independent testing.
INTRODUCTION
- Harmonized technical requirements and effective assessment procedures are essential to consistent automated-driving safety, with UNECE regulations and approval authorities establishing common requirements across markets.The effectiveness of this framework depends on whether prescribed tests capture the evaluated systems’ operational behavior and safety limitations.Source: INTRODUCTION · p. 3
- Proving-ground tests are repeatable and reproducible but cover limited scenarios that may not represent real-world driving complexity or system behavior across public-road operating domains.ALC geo-fencing can require manufacturer collaboration for proving-ground tests, creating a barrier to independent market surveillance; limited RTK GNSS availability also complicates accurate testing in tunnels, canyons, and rural areas.Source: INTRODUCTION · p. 3
- The campaign coordinated a Vehicle Under Test and an approaching Take-Over vehicle on French motorways using a full factorial test matrix and LiDAR, cameras, GNSS receivers, and dedicated storage equipment.LiDAR processing measured inter-vehicle distances and relative velocities and compared observed trajectories with UNECE R79 lane-change suppression requirements; RTK GNSS served as ground truth for LiDAR precision and detection rate and for assessing statistical significance.Source: INTRODUCTION · p. 3
- The paper’s main contribution is a methodology for independently testing ALC systems on public roads using LiDAR measurements rather than conventional proving-ground procedures and RTK GNSS-based instrumentation.The authors state that the approach can support type approval and market surveillance under realistic driving conditions more easily, efficiently, and economically than conventional approaches.Source: INTRODUCTION · p. 3
Related Works
Prior work covers ALC development and field evaluation, broader driver and lane-change research, and LiDAR-based traffic safety assessment. However, the literature lacks a public-road test of production ALC behaviour against UN Regulation No. 79 using vehicle-mounted LiDAR, including system-suppressed attempts.
ALC studies
- Ulbrich et al. (2017) studied scenario-based tactical lane-change planning, Raboy et al. (2021) demonstrated cooperative ALC in a field experiment, and Mattas et al. (2025) evaluated five commercial systems but omitted suppressed manoeuvres.Mattas et al. compared observed margins for three vehicles with UN Regulation No. 171 using GNSS/IMU and road-camera data.Source: Related Works · p. 3
Driver behaviour and models
- Noonan et al. (2022) analysed driver-initiated Tesla Auto Lane Changes without measuring surrounding-traffic interaction; Madigan et al. (2018) studied overtaking in a simulator, and Schakel et al. (2012) modelled human lane-changing and car-following.Source: Related Works · pp. 3–4
Reviews and on-road ADAS testing
- Donà et al. (2026) extended a simulation safety model for UN Regulation No. 157 scenarios but did not test production ALC; reviews likewise lack a dedicated UN Regulation No. 79 ALC procedure.On-road ADAS studies have predominantly addressed longitudinal control, including ACC interaction and system assessments.Source: Related Works · p. 4
LiDAR assessment
- LiDAR studies use roadside sensors for vehicle and vulnerable-road-user tracking, with scalability limits, or mobile sensors for broader safety assessment rather than type approval or market surveillance.Examples include Bhattarai et al. (2024), Gargoum et al. (2022), Wu et al. (2020), Kilani et al. (2021), Guerrero-Sevilla et al. (2025), and Kim et al. (2025).Source: Related Works · p. 4
Research gap
- The literature lacks a public-road campaign testing ALC against UN Regulation No. 79 with vehicle-mounted LiDAR, and prior work does not study system-suppressed attempts.Existing work considers only completed ALC manoeuvres.Source: Related Works · p. 4
METHODS
The method tests whether commercial UNECE Regulation No. 79 type-approved Assisted Lane Change systems suppress public-road manoeuvres when the approaching vehicle is closer than the regulation’s Critical Distance. Critical Distance is evaluated when the test vehicle crosses the lane marking.
Critical Distance for Assisted Lane Change Manoeuvres
- The Critical Distance depends on the speeds of the approaching and Assisted Lane Change vehicles and uses 3 m/s^2 deceleration, 0.4 s reaction time, and a 1 s resulting time gap.The supplied passage defines these parameters but does not provide a complete readable expression for the distance.Source: METHODS · p. 4
- The regulation evaluates Critical Distance when the vehicle under test crosses the lane marking; if the approaching vehicle is closer, the system should suppress the manoeuvre.This is the lane-change procedure’s Time of Start Measure, as illustrated in Figure 1.Source: METHODS · p. 5, Figure 1 · p. 5
- The study investigates whether type-approved systems reject public-road manoeuvres initiated with the approaching vehicle below R79’s Critical Distance.The Critical Distance is the required separation between the vehicle under test and the take-over vehicle for an Assisted Lane Change to start.Source: METHODS · pp. 4–5
Experimental Setup and Testing Procedure
The study tested a category C Assisted Lane Change system on the A31 motorway using a three-vehicle procedure and LiDAR/GNSS instrumentation. Of 27 manoeuvres, 18 completed and 9 were suppressed, including completed manoeuvres below the R79 critical-distance threshold.
Experimental Setup and Testing Procedure
- In March 2026, three type-approved passenger cars tested a category C R79 system on the northbound A31 between Dijon and Nancy.The VUT’s assisted lane-change function was checked before the campaign and confirmed to operate on geo-fenced highway stretches.Source: Experimental Setup and Testing Procedure · p. 5
- The support car maintained a fixed distance behind the VUT and served as a marker for triggering ALC as the faster TO approached in the adjacent lane.At the Time Start Procedure, the TO aligned with the support, and the VUT activated its indicator and moved toward the lane marking.Source: Experimental Setup and Testing Procedure · p. 5, Figure 2 · p. 6
- The VUT used a commercial LiDAR vehicle-detection system with GNSS, while the TO and support cars carried GNSS/INS receivers for ground-truth positioning.The system provided ego and tracked-object poses at 10 Hz, including position, velocity, heading, width, and length; dashboard and outside cameras supported manual annotation of indicator activation and lane crossing to one-second resolution.Source: Experimental Setup and Testing Procedure · p. 6
- For each VUT–TO speed combination, testing triggered ALC at three distances selected around the R79 critical and minimum activation distances.Some critical combinations were repeated; speeds were constrained by the motorway’s 130 km/h limit. Tests covered 100–130 km/h and trigger distances of 20–60 m.Source: Experimental Setup and Testing Procedure · pp. 6–7
- Although support-car triggering was more repeatable than manual driving, variable activation delay and TO speed changes limited measurement of the manoeuvres.The TO’s adaptive cruise control reacted to VUT lateral motion, and three completed manoeuvres were outside LiDAR range and therefore untracked.Source: Experimental Setup and Testing Procedure · p. 7
- Of 27 ALC manoeuvres, 18 completed and 9 were suppressed; 33% of completed manoeuvres were initiated below the R79 Critical Distance at lane crossing.Six completed cases were identified as potential Critical Distance overshoots; Figure 4 compares trajectories with the R79 threshold.Source: Experimental Setup and Testing Procedure · pp. 7, 9, Figure 3 · p. 7, Figure 4 · p. 8
DISCUSSION
The discussion finds three public-road lane-change overshoots significant at 99% confidence, while explaining how LiDAR uncertainty, low speed differences, and the limited scope of UNECE R79 approval tests shape the assessment.
DISCUSSION
- LiDAR VDT position precision was 0.83 m (std) and velocity precision was 1.40 km/h (std), estimated against RTK GNSS during intervals with RTK Fixed integer-level precision.Simultaneous RTK Integer Fix was available for both vehicles in only 2.3% of recorded data; manual annotation of lane-marking crossing had 1-second resolution.Source: DISCUSSION · p. 9
- 3 of the six identified critical-distance overshoots (experiments 16, 16b, and 16d) had 3-sigma or 99% confidence; experiments 10, 10b, and 10d had only 2-sigma significance.The final categorization identifies the three 99%-confidence overshoots as a subset of Critical Completed experiments.Source: DISCUSSION · p. 9, Figure 5 · p. 10
- The authors conclude that LiDAR-based measurements can detect R79 Critical Distance Overshoots on public roads and are fit for type approval.The assessment uses RTK GNSS as ground truth, although LiDAR measurements were more available and less precise because of VDT pipeline hypotheses and approximations.Source: DISCUSSION · p. 9
- All six unsuppressed Critical Distance Overshoots occurred at low delta-v, where the R79 curve is flat, but the 1-second time-headway limit still applies.The rule assumes the take-over vehicle is faster than the VUT and limits required deceleration; the authors identify low delta-v as a case where critical-threshold detection may fail.Source: DISCUSSION · pp. 9–10
- The overshoots occurred outside the operating domain prescribed for UNECE R79 type-approval tests, which require one critical manoeuvre abort test on a proving ground below 100 km/h.Testing personnel did not perceive any Critical Completed manoeuvre as dangerous, but the authors argue that systems should be robust to the combinations encountered on public roads.Source: DISCUSSION · p. 10
CONCLUSIONS
The study demonstrates a LiDAR-based methodology for independent public-road assessment of type-approved Assisted Lane Change systems, and identifies completed manoeuvres that overshot UNECE Regulation No. 79’s critical-distance threshold.
CONCLUSIONS
- The methodology uses LiDAR-based measurements instead of conventional RTK-GNSS instrumentation for independent public-road testing of type-approved Assisted Lane Change systems, including geo-fenced systems, without manufacturer support.It enables coordinated testing without relying on high-precision GNSS data.Source: CONCLUSIONS · p. 10
- 6 of 27 experiments completed despite an R79 Critical Distance Overshot; 3 of those 6 had a 99% Confidence Interval measurement of the overshot.The experiments covered different relative distances and speeds in a full factorial design.Source: CONCLUSIONS · p. 10
- Public-road LiDAR testing was precise enough to detect critical system behaviour in real-world operating regions that conventional type-approval testing may insufficiently represent, supporting independent market surveillance and in-service monitoring.Source: CONCLUSIONS · p. 10
- Testing was limited to one vehicle and measurement and temporal-annotation precision; future work targets LiDAR accuracy, multisensor fusion, and automated event detection.These developments are intended to support a repeatable procedure for market surveillance, in-service monitoring, and evaluation of future R79 revisions.Source: CONCLUSIONS · p. 10
FUNDING
The authors disclosed no financial support for the research, authorship, or publication of the article.
FUNDING
- The authors disclosed no financial support for the research, authorship, or publication of the article.Source: FUNDING · p. 11