LDW and LKA systems are becoming more prevalent on modern vehicles, however their current simple implementation leads to frequent activation and intervention. Drivers report this indiscriminate intervention as annoying and perceive it to be unnecessary, which leads to system deactivation and the loss of any potential safety benefit in critical situations.

This paper focuses on developments relating to actively intervening Lane Support Systems (LSS), namely Emergency Lane Keeping (ELK) systems. It describes the findings of analyses of relevant European real world collision data and the development of representative test scenarios, methodology and evaluation criteria.

The aim of developing ELK test methodology is to encourage LSS that intervene less frequently but more effectively to prevent collisions occurring in critical lateral control situations. This was promoted by linking intervention to the associated threat thus encouraging vehicle technology that discriminates between event types and their criticality. A benefit of interpreting the threat is the ability to minimise the frequency of corrective inputs thus reducing the driver perception of unnecessary intervention and maximising driver acceptance of such systems. Therefore, similar to Autonomous Emergency Braking (AEB), the demanding requirement for a system to be default on at the beginning of every journey can be made to maximise the potential benefit in every day driving

Relevant real world crash data from European sources was studied to identify the common features of lateral control collisions such as run-off road, head-on collisions with oncoming vehicles and collisions with overtaking vehicles. Test scenarios, methodology, metrics and controls were developed for evaluating ELK systems. Evaluation criteria were also developed to encourage effective intervention in the real world.

The oncoming and overtaking testing scenarios require the use of a partner vehicle to trigger the ELK system intervention. The test target specified for use is the Global Vehicle Target (GVT), an impactable 3D car target according to ISO 19602 Part 1. It was intended that a representative road edge would also be developed to achieve repeatable testing across various locations, however it proved more challenging than initially anticipated to replicate the attributes of a real road edge and work continues in this area.

Although the testing methodology is grounded in real world data, certain compromises were required relating to the steering input and how the lane departure was generated in order to achieve a repeatable and reproducible results. However the results of initial testing indicate that it is possible to differentiate between system performances, particularly at higher lateral velocities.

The test procedures have been proposed to Euro NCAP for adoption, and will be implemented in the star rating scheme from 2018. Integration of the procedures into consumer testing will help to guide development of system design in future vehicles.