Automated driving in lane merge scenarios


The next step towards the big dream

No longer a mere pipedream, automated cars are increasingly becoming reality. However, despite all the rapid progress that has been made in recent years, the ultimate goal of fully automated driving – which would allow the driver to sit back, relax and enjoy the environment – is still far-off. This is where Valeo, one of the leading companies in smart mobility, steps in. To gain new ideas based on a different perspective Valeo brought in a group of TU/e PDEng trainees to investigate in one of the core parts of automated driving, in the context of a 8 weeks in-house training project.

The group consisted of a multidisciplinary team with 11 trainees from Software Technology, Mechatronic Systems Design and Automotive System Design. It was precisely this mix that made this team so strong, bringing together perspectives from various disciplines and training courses. The trainees themselves also recognize that: “Working in a multidisciplinary team was a wonderful and educational experience”.

Valeo is an automotive supplier and partner to automakers worldwide. As a technology company, they design innovative solutions for smart mobility, with a particular focus on intuitive driving and reducing CO2 emissions. Self-driving cars can be distinguished according to five levels, with each level representing a further step in autonomy. Valeo was one of the first companies introducing level 2 systems to the market, namely remote parking, where the driver can be even outside of the car. Currently, the supplier conducts research and development towards higher levels of automation.

Black box

The PDEng assignment was related to the automatic lane change strategy in different situations: when the car has to change lanes either because the roadway stops (for example with a highway entry) or two roadways become one, a lane merge. The solution had to be based on Valeo’s specific system engineering methodology when it comes to automated driving solutions. The following quality requirements had also to be taken into account: availability, testability, compatibility, safety and comfort.

The trainees first looked at the interaction with other systems in the car. The big question here was: what should this specific lane merging system do and on what should it react? Filling the black box, so to speak. All kinds of available information – such as lines, lanes, road markings, road signs, merging, acceleration, lane length and highway speed – were considered and used if applicable.

An existing simulation was then extended to see if the ideas are working as expected. Issues such as ‘How does the strategy change based on your speed?’ and ‘How should the vehicle react on different speeds of other cars?’, have all been brought up before the simulation and been verified in different simulations.

These simulations consist of two parts: before changing lanes and afterwards. Finally, these two simulations were combined to get a full picture. The trainees faced a variety of challenges both in technical and process aspects, and above all, working in a multidisciplinary and multicultural team together. They managed to overcome the challenges and improved their skills during this training project.

Reliable system

The trainees have successfully completed the training project and have integrated basic parts of their proposals into the existing solution and thereby extended the algorithms. They conclude that it can certainly become a reliable system for use in the car when integrated with navigation and other systems, such as adaptive cruise control.

In addition, they indicate that the behavior of the feature is not yet reliable under all conditions. For example, it now only takes into account other cars as other road users. “Motorcyclists still have to be added”, the group indicates. “They are also part of the traffic. Without these kind of additions the system is not complete and doesn’t work optimal.”

Valeo is pleased with the work that the group of trainees have done and the way in which Valeo’s proprietary system engineering methodology has been applied successfully: “The results of the project will certainly take us one step further and gave us good impression where to look closer. New questions appeared in the end and some proposals will be worthy to be investigated in more detail. So that might be something for the next group of PDEng trainees.”

PDEng ST/ASD/MSD programs

The two-year salaried post-Master designer programs Software Technology (ST), Automotive Systems Design (ASD) and Mechatronics Systems Design (MSD) prepare trainees for a career in the high-tech industry as technological designer, and later on as a software or system architect. The programs consist of a number of modules dedicated to specific parts of the underlying scientific fields and technology domains. One of these modules is devoted to Software Embedded Systems, which plays an increasingly important role in both Automotive and Mechatronics. The programs are concluded with a major design project of 10-12 months in a company.