In the automotive domain, the following use cases are defined:


Highly automated driving on single- or multi-lane highways is one of the first applications for passenger cars as well as commercial vehicles. It allows the driver to turn attention away from the driving task for a longer period. The vehicle handles all of the driving-related tasks like overtaking slower vehicles, driving in tunnels or through toll booths. In addition, the system initiates the process of returning responsibility back to the driver. In emergency situations the functions execute defined actions. The use case “Highway Pilot” focuses on:

  • Description of test requirements of vehicle functions for 24/7 usage at any possible environmental condition like bright/dark, hot/cold, dry/wet
  • Semi-virtual validation of sensor performance and algorithms in a real vehicle
  • Seamless validation tool chain from Model-in-the loop (MiL) to Vehicle-in the-loop (ViL) following the ENABLE-S3 architecture
  • Generation of test scenarios for highway pilot
  • Comparability of tests in different test environments (MiL, ViL, Driver-in-the-loop)

The DrivingCube is an automotive test bed (power train test bed or alternatively a chassis dynamometer) extended by physical sensor stimulators. It is used to reproduce plausible, physically correct environmental conditions for the vehicle and therefore for the automated driving function. The test bed includes dynos, vehicle fixation and hardware interfaces for the sensor stimulators. In order to facilitate the development and facilitate the demonstration, there is also a small-scale 1:10 version of the test bed.  The intention of this demonstrator is to integrate the different tools and methods, related to the Highway Pilot testing functions that have been developed during the project. The car is equipped with an automated highway pilot function as well as the required sensors. The considered sensors will be camera, radar, and ultrasonic (just used for the small 1:10 scale version of the driving cube). The environment is simulated in Vires VTD.


The crossing of a road intersection is difficult and might lead to hazardous situations. In urban areas, almost 50% of traffic accidents with physical injuries happen within intersections and branches. Left turns at intersections are particularly challenging because of the large number of factors, which must be considered. An important factor is the surrounding traffic  especially the vehicles on the oncoming lanes, which should be crossed during a left turn. This use case focuses on:

  • developing processes, tools and experimental trials to validate decision-making functions
  • considering theoretical frameworks, using digital models, and executing a limited number of field trials
  • determining all likely external and internal disturbances autonomous systems might encounter in real traffic conditions
  • demonstrating safe intersection crossing, validation & verification over a defined scenario set

As a first step towards validation of UC2, a preliminary experiment was conducted in 2017 both in simulation and reality (i.e. proving ground). The same intersection, vehicles and sensors were used for both environments. Although the left turning System Under Test (SUT) was partially implemented (i.e. human was still driving SAE0), the preliminary experiment was conducted to fine tuning scenarios, define best Key Performance Indicators (KPIs) and to detect potential problems as well as accumulate experience before the second, main experiment this year.

First, several variations of scenarios were introduced in order to challenge the automation on different levels. The focus was to optimize vehicles’ parameters (e.g., starting position, speed, acceleration) to create scenarios V&V of the SUT relevant key senses and situations. As a result, four basic scenarios and nine parameters were identified and tested.

Together with project partners, a list of KPIs was derived from the safety requirements of Use Case 2. The KPIs can be divided into system and human related KPIs.

Finally, data were collected from four participants both in reality and simulation. The collected data were analyzed to define methods to compare SUT within the two environments and to test the interpretation of the defined KPIs. A first results from proving ground revealed that the intersection can be divided into different areas of interest based on the different phases of turning manoeuver. The areas were highlighted in the analysis to enhance data interpretation.

The trajectories of the ego vehicle collected from the different scenarios were plotted on the intersection layout. This helps to understand the path choice of human driver and to identify stop positions on the intersection as baseline.

The analysis of the collected data from the preliminary experiment is still an ongoing.

Additionally, several collaborations regarding V&V of C2X communication used by an autonomous left turning vehicle take place within UC2. In cooperation with AIT, DLR performed a measurement campaign on proving ground (as baseline) and in simulator to evaluate a real-time C2X communication model developed by AIT. Data on C2X signal strength, packet drop rates and further parameters were collected in both environments. This enables a comparison of the C2X communication in both test environments and allows a validation and optimization of the AIT real-time C2X model.

Further, in cooperation with GUT DLR performed another measurement campaign on the proving ground. In this campaign a Jammer device from GUT interrupts the C2X communication of the DLR test vehicle. The collected data will help to develop a Jammer model. With this model, C2X signal, Jamming will be already possible to test in a simulated V&V environment. The following video shows the jamming setup on the proving ground.




The main objective of this Use Case is the development of the cognitive engine for the in-car reasoning system. The benchmarks for testing the system will be based on data from Dublin, Ireland, comprising traffic volumes, speeds, accidents, intersection cycle time, and major events held in the area, together with data from social media and news feed. Further, data on route preferences and re-routing preferences will allow for the evaluation of recommender systems. Moreover, it is the goal to develop and test signalling strategies for achieving balancing of traffic loads.

We consider the connected car scenario and aim to push the limits of off-line testing as far as possible within this scenario. In our use case, a context-aware in-car reasoning system supports passengers in their travel decisions. For instance, in the context of a motorway ride, assume that a large traffic accident occurs on the planned route. The reasoning system would detect that the accident will affect the arrival time at the destination, alert the driver and provide, upon request, alternative routes specifying the consequences in terms of cost, average arrival time and uncertainty in arrival time. It would also provide assistance to the driver in choosing among the proposed alternatives by asking information about preferences (for instance preferring to take a toll road to ensure a high probability of arriving at a given time). To this end the in-car reasoning system must have a contextual understanding of the vehicle’s surroundings and be able to interact in a meaningful manner with the driver. This interaction can involve advice and discussion of route options up to context aware alerts issued to the driver.

IBM Research - Ireland and UCD are currently working on in-depth validation of the Context-Aware In-Car Reasoning System by utilising Hardware-In-the-Loop platform that enables embedding a proof-of-concept real vehicle, equipped with the system, into an emulated large-scale traffic scenario. The platform thus permits a real vehicle and driver using the Context-Aware In-Car Companion to interact with thousands of simulated cars on a common road map, in real-time. Various aspects of such interaction were tested by the following demonstrators:

1) Parsing Engine illustrates the interaction of the Context-Aware In-Car Companion with social media (e.g. Twitter)

2) Cognitive distraction classifier demonstrates how the Companion may react to the current cognitive driver state

3) Personalised Pedestrian Alert demonstrates how the Companion can warn pedestrians about electric vehicles which are predicted to approach a pedestrian crossing

This picture shows how the car is departing from the IBM Research building and the Companion in the car broadcasts its predicted destination, so that a specially developed app on a pedestrian's phone can generate warnings.


This Use Case is intended to allow vehicles to safely drive their passengers through high-density traffic at speeds up to 50-kph without driver input by using a mixture of adaptive cruise control and lane-keep assist systems. In the proposed approach, this system will be enhanced with information coming through V2X communications, eventually allowing the creation of vehicle platoons. The main research topics are based on the development and validation of safety mechanisms to be used by automotive V2X applications and preventing hacking as well as impersonation of “V2X-enabled” systems.

The Traffic Jam Pilot demonstrators have recently entered an end-to-end validation process to confirm readiness for their final trials.
The tasks ahead involve data collection for the evaluation of results and KPIs from the demonstrators.
The first demonstrator implemented a simulated environment with the autonomous function running on a set of vehicular robots, capable of communicating with each other, and cooperating either via the traffic-jam-assist or the platooning function. Special focus will be put on communication scenarios such as malicious fault injection.
The second demonstrator runs on the same functions, but is centred on the study of the human driver behaviour. Human trials are expected to begin before December.



Valet Parking is expected to be the first commercially available automated driving function at SAE Level 4. In this Use Case, methods and tools are explored for developing a safe automated parking system. This includes:

  • the specification of a test system that allows automated scenario generation and execution
  • distributed system under test
    that incorporates in-vehicle and remote functional components in the Parking Area Manager (PAM)
  • functional safety concept concerning the implications of safety regulations and relevant existing standard
  • test coverage analysis
  • perception models of different sensor types