Evaluation of driver assistance functions in new energy vehicles (NEVs) is a systematic verification process that confirms the safety, performance, and regulatory compliance of Advanced Driver Assistance Systems (ADAS) such as adaptive cruise control, automatic emergency braking, and lane keeping assist. This article outlines the applicable international and national standards, test sample categories, evaluation items, laboratory and proving-ground test methods, and typical application scenarios. It serves as a practical technical reference for vehicle manufacturers, system suppliers, and engineers planning accredited third-party evaluation for market access and product validation.
Technical Overview
Modern new energy vehicles, including battery electric vehicles (BEVs), plug-in hybrid electric vehicles (PHEVs), and fuel cell electric vehicles (FCEVs), are increasingly equipped with Level 2 and Level 2+ driver assistance functions. These systems combine perception sensors, domain controllers, and chassis-by-wire actuators to deliver functions such as adaptive cruise control (ACC), automatic emergency braking (AEB), lane keeping assist (LKA), blind spot detection (BSD), and automated parking assist (APA). Because these functions directly influence active safety and interact with the electric powertrain, regenerative braking, and high-voltage architecture, a rigorous and standardized evaluation is essential before market release.
The evaluation of driver assistance functions for NEVs is built on four pillars: functional performance testing against measurable key performance indicators (KPIs), safety assurance in accordance with ISO 26262 (functional safety) and ISO 21448 (Safety of the Intended Functionality, SOTIF), cybersecurity validation aligned with ISO/SAE 21434, and environmental robustness verification covering EMC, weather, lighting, and sensor contamination conditions.
Principal applicable standards and regulations include:
- ISO standards: ISO 15622 (ACC), ISO 15623 (forward collision warning), ISO 17387 (lane departure warning systems), ISO 17386 (low-speed maneuvering aids), ISO 22839 (forward vehicle collision mitigation), ISO 21979 (emergency steering support), ISO 16505 (camera monitor systems), ISO 26262, ISO 21448, and ISO/SAE 21434.
- UN regulations: UN R130 (lane departure warning), UN R131 and UN R152 (automatic emergency braking), UN R140 (electronic stability control), UN R151 (blind spot information), UN R159 (moving-off information), UN R79 (steering equipment and automatically commanded steering functions), UN R156 (software update management), UN R155 (cybersecurity), and UN R160 (event data recorder).
- Chinese national standards (GB): GB/T 39263 (ADAS terminology), GB/T 39323 (lane keeping assist for passenger cars), GB/T 39901 (AEBS for passenger cars), GB/T 41797 (automated parking assist test methods), GB/T 38186 (AEBS for commercial vehicles), and the mandatory intelligent connected vehicle standards covering cybersecurity, software upgrade, and data recording.
- Regional regulations: EU General Safety Regulation requirements (intelligent speed assistance, driver drowsiness warning, AEB, and event data recording) and FMVSS 127 (AEB for light vehicles in North America).
Accredited third-party laboratories and proving grounds typically structure the evaluation as a staged program: component and subsystem verification, closed-track scenario testing, simulation-based scenario coverage analysis, and finally real-world road validation.
Testing Samples
Complete Vehicles
Production-intent or production NEVs in full trim configuration are the primary sample for system-level evaluation. The vehicle sample must reflect the final sensor layout, software calibration, tire specification, and mass distribution. Electric vehicles are often tested at multiple state-of-charge levels, because battery mass and regenerative braking behavior can influence braking and cruising performance.
ADAS Sensor Modules and Kits
Individual perception components, including front cameras, surround-view cameras, millimeter-wave radars (77/79 GHz), 4D imaging radars, LiDAR units, and ultrasonic sensors, are evaluated as standalone samples for detection range, field of view, angular resolution, classification accuracy, and environmental durability.
Domain Controllers and Electronic Control Units
Fusion controllers, ADAS domain controllers, and associated ECUs are tested for signal processing latency, algorithm output accuracy, diagnostic coverage, electrical characteristics, and resilience to voltage transients in the 12 V/48 V and high-voltage vehicle electrical environment.
Actuator Subsystems
Brake actuators (including electro-hydraulic and electro-mechanical brake-by-wire systems), electric power steering racks, and torque vectoring units are evaluated for response time, control precision, and fail-safe behavior, since assistance functions ultimately depend on actuator performance.
Software Releases and OTA Packages
Specific software builds, calibration datasets, and over-the-air (OTA) update packages are treated as test samples. Each release is subject to regression testing to confirm that updated perception, planning, or control logic has not degraded previously validated functions.
Testing Items
The evaluation scope covers function-level performance, safety assurance, and NEV-specific integration behavior. Typical testing items include:
- Adaptive cruise control (ACC): target acquisition range, set-speed control accuracy, time-headway maintenance, smoothness of acceleration and deceleration, stop-and-go behavior, and cut-in vehicle response.
- Automatic emergency braking (AEB): collision avoidance and mitigation against stationary, moving, and braking vehicle targets (CCRs, CCRm, CCRb scenarios), adult and child pedestrian scenarios including night conditions (CPNA, CPLA, CPNC), and cyclist scenarios, measuring speed reduction, braking onset timing, and time-to-collision margins.
- Lane keeping and lane departure functions: warning timing, lateral offset correction capability, curve negotiation, lane marking recognition across colors and wear states, hands-off detection, and emergency lane keeping (ELK) intervention against roadside and oncoming hazards.
- Blind spot and rear-area functions: blind spot detection range and timing, rear cross-traffic alert and braking, door open warning, and rear collision warning.
- Automated parking assist (APA) and remote parking (RPA): parking slot detection rate, path planning success, parking accuracy, obstacle reaction during maneuver, and remote control latency and safety stop behavior.
- Traffic jam and integrated cruise assistance (TJA/ICA): low-speed following stability, stationary object handling in congestion, and transition logic between longitudinal and lateral assistance.
- Driver monitoring and warning functions: drowsiness and distraction detection accuracy, warning effectiveness, and intelligent speed assistance (ISA) sign recognition accuracy.
- Safety assurance items: functional safety audit and testing to ISO 26262, SOTIF residual and unknown-hazard analysis per ISO 21448, and cybersecurity testing aligned with ISO/SAE 21434.
- NEV-specific integration items: blending of regenerative and friction braking during assistance interventions, creep and hill-hold interaction, drive-mode-dependent system behavior, interaction with pedestrian acoustic warning systems, and range impact of assistance functions.
- Human-machine interface (HMI): warning modality (visual, acoustic, haptic), timing, salience, comprehensibility, and mode transition feedback.
- Environmental robustness: performance under low sun angle, night, rain, fog, tunnel entry and exit, sensor contamination (dirt, water film, snow coverage), and electromagnetic compatibility per CISPR 25, ISO 11452 series, ISO 7637 series, and ISO 10605.
Testing Methods
Evaluation programs combine physical testing, simulation, and road validation to achieve both regulatory conformity and sufficient scenario coverage.
- Closed-track scenario testing: Vehicles are tested on instrumented proving grounds using crashable soft vehicle targets mounted on self-propelled platforms, articulated pedestrian and cyclist dummies, and driving robots that control steering, throttle, and braking with high repeatability. Ground-truth data is captured with GNSS-RTK positioning, inertial measurement units, and optical speed and distance measurement at centimeter-level accuracy.
- KPI-based quantitative assessment: Measured quantities include time-to-collision at intervention, achieved speed reduction, braking or steering onset latency, lateral deviation from lane center, minimum following distance, and warning lead time. Results are scored against pass/fail thresholds defined by the applicable standard or rating protocol.
- Scenario-based safety evaluation: In line with scenario-based evaluation frameworks such as ISO 34502, test catalogs are derived from traffic accident databases, naturalistic driving data, and hazard analysis to cover critical scenarios including cut-ins, cut-outs, crossing pedestrians at occlusion, and adverse weather edge cases.
- Simulation and X-in-the-loop methods: Model-in-the-loop (MIL), software-in-the-loop (SIL), hardware-in-the-loop (HIL), and vehicle-in-the-loop (VIL) testing are used to extend scenario coverage far beyond physical track limits. Standardized road network and scenario description formats such as OpenDRIVE and OpenSCENARIO, together with validated sensor models, enable virtual verification of perception, planning, and control stacks.
- Laboratory environmental and EMC testing: Sensor units and controllers undergo climatic and mechanical stress testing based on the IEC 60068 series, vibration profiles, and radiated and conducted EMC immunity testing per CISPR 25 and the ISO 11452 series to verify performance in the high-voltage NEV electrical environment.
- Real-world road validation: Structured on-road trials on highways, urban arterials, and complex intersections validate performance under mixed traffic, diverse lane markings, V2X signal environments, and varying lighting. Data loggers record multi-channel sensor, decision, and actuation signals for post-processing and event data review in accordance with EDR requirements of UN R160 and corresponding GB requirements.
- OTA and regression testing: Software releases are verified through a defined regression matrix, delta-risk analysis, and re-execution of critical scenarios to confirm continued compliance following updates, consistent with software update management requirements under UN R156.
Applications
Evaluation results of driver assistance functions in new energy vehicles support a wide range of engineering and commercial objectives:
- Regulatory type approval and market access: Demonstrating conformity with UN regulations, GB standards, and regional requirements such as the EU General Safety Regulation and FMVSS 127 for product homologation, import, and sales authorization in target markets.
- Compulsory certification and announcement testing: Meeting mandatory national requirements for vehicles sold in China, including applicable ADAS, cybersecurity, software upgrade, and data recording provisions.
- Consumer safety rating programs: Preparing vehicles for new car assessment program protocols, which weight AEB, lane support, speed assistance, and night pedestrian protection heavily in overall star ratings.
- R&D validation and design verification: Benchmarking perception and control performance, calibrating actuator blending on electrified platforms, and supporting design verification (DV) and production validation (PV) milestones.
- Supplier quality and SOP release: Objective evidence of subsystem performance for tiered supply chains, sample release, and production part approval processes.
- OTA lifecycle management: Verifying that continuous software improvements maintain certified safety performance throughout the vehicle service life.
- Insurance, fleet, and export compliance: Providing documented safety performance data for insurance risk modeling, fleet procurement, and cross-border export documentation.
FAQ
What is the difference between evaluating a driver assistance function and an automated driving function?
Driver assistance functions (typically SAE Level 1 to Level 2) require continuous driver supervision, so evaluation focuses on intervention quality, warning effectiveness, and driver monitoring. Automated driving functions assume the system is responsible within an operational design domain, which adds requirements for fallback behavior, minimal risk maneuvers, and more extensive SOTIF and scenario coverage evidence.
Which standards are most commonly referenced for ADAS evaluation?
The core references are ISO 15622, ISO 15623, ISO 17387, ISO 17386, and ISO 22839 for individual functions, UN R130, R131, R151, R152, and R159 for regulatory conformity, ISO 26262 and ISO 21448 for safety assurance, plus GB/T 39323, GB/T 39901, and GB/T 41797 for the Chinese market. The exact list depends on the target market and the functions installed on the vehicle.
How is AEB performance quantified?
AEB is quantified by the collision avoidance speed (the maximum test speed at which the system fully avoids impact), achieved speed reduction in unavoidable scenarios, time-to-collision at first warning and at braking onset, and false-positive rate across non-hazardous scenarios, measured against standardized vehicle, pedestrian, and cyclist target scenarios.
Why do new energy vehicles require specific evaluation considerations?
Because electric powertrains deliver instant torque, strong regenerative braking, and near-silent operation, assistance functions must be evaluated for brake blending accuracy, deceleration smoothness, mode-dependent behavior, interaction with battery management under repeated high-deceleration events, and performance within a more electromagnetically complex high-voltage environment.
How long does a complete evaluation typically take?
Timeline depends on scope. A single-function regulatory test may be completed within one to two weeks on track. A comprehensive program covering multiple functions, environmental robustness, simulation coverage, and road validation typically spans two to six months, including reporting and any corrective-action retesting.
Do OTA software updates require re-evaluation?
Yes. Software changes that affect perception, planning, control, or safety-relevant behavior require regression testing and delta-risk assessment. Regulatory frameworks such as UN R156 and the corresponding GB software upgrade requirements mandate documented management of such updates throughout the vehicle lifecycle.
Can simulation results replace physical track testing?
Simulation substantially extends scenario coverage, especially for rare and dangerous edge cases, and X-in-the-loop methods are increasingly accepted as supporting evidence. However, regulatory type approval and rating protocols still require physical proving-ground verification of standardized scenarios using calibrated targets and ground-truth instrumentation, typically performed at accredited third-party laboratories and certified test sites.