Overview: BMW’s automated driving research

Motorway A9 from Munich to Nuremberg. As usual there is a high volume of traffic. But despite the stress of the situation, the driver sits calm and relaxed behind the wheel. Why? Because the car is driving itself. It brakes, accelerates and passes other vehicles on its own, while monitoring and adapting to the prevailing traffic conditions.

The other reason is that the driver is Dr. Nico Kämpchen, Project Manager of Highly Automated Driving at BMW Group Research and Technology. Kämpchen has already completed nearly 5000 test kilometres. To those of us who feel instinctively compelled to grip the wheel even when driving at a snail’s pace in traffic, 5000km sounds like a fair time to get used to leaving the car to its own devices at high speed.

Automated BMW 3-series.

BMW Group Research and Technology has been working for several years on the development of electronic co-pilots to support automated driving in specific situations — for example, the BMW Track Trainer, tested on the race track, as well as adaptive cruise control (ACC) and the Emergency Stop Assistant. To understand further the potential offered by these systems, as well as their limitations, researchers are ready to take their next and rather scary step: developing advanced driver assistance systems for the motorway.

Researchers equipped a 5-series saloon with control software, vision assistance equipment and environment detection systems. The automated assistance function for motorway journeys can be activated and deactivated by the driver.

Once switched on, the prototype system can autonomously control acceleration and braking, and it can safely pass slower vehicles. One of the biggest problems early in the project was reacting to vehicles merging onto the motorway. The prototype system reacts to the situation by allowing the merging vehicles to join the traffic flow, and it can even change lanes giving the merging vehicles adequate space to enter traffic safely. This is possible up to a speed of 130km/h, but in compliance with current traffic regulations regarding speed limits and such things as prohibited passing zones.

‘This is an entirely new situation and experience for the driver. It is a strange feeling handing over complete control of the car to an autonomous system. But after a few minutes ... drivers and passengers begin to relax somewhat and trust the independent system,’ says Nico Kämpchen. ‘Nevertheless, the driver is still responsible for the situation at all times and must constantly keep an eye on traffic and the surroundings.’

To make sure that the automated research vehicle functions effectively in real traffic, the car must have algorithms to deal with daily traffic situations. The basis for these strategies comprises two parts: first, pinpointing the position of the vehicle in its own lane is essential; and second, the car must be able to recognise all vehicles and objects in its nearest surroundings. This is achieved through the redundant fusion of various sensor technologies such as lidar, radar, ultrasound and video cameras that monitor the environment around the automobile. (Redundant in the computer technology context of having more processing power than you ‘need’ — the definition of what you ‘need’ being rather vague.)

To ensure that the vehicle’s situation is precisely assessed, at least two different measurement methods have to be used in every direction. In this way, engineers can be sure that a potential weakness in one method is ‘caught’ by another system.

Using digital maps, the camera and the localisation data of the GPS, the automated vehicle prototype can determine its position in its lane. It also receives exact information about the characteristics of the route ahead, including the number of lanes available. This information is supplemented by data from the forward-looking camera integrated in the lane departure warning system. Objects in front of the vehicle are detected by the radar sensors of the adaptive cruise control system and by a laser scanner as well. The same is true for objects at the sides or rear of the vehicle.

The new technology needed for automated driving assistance systems was developed by engineers at BMW Research and Technology in two pioneering projects that produced the BMW Track Trainer and the Emergency Stop Assistant.

The BMW Track Trainer supports autonomous driving on competition race courses. A very dynamic driving style can be adopted safely. The Track Trainer is currently used in BMW driver training sessions to give participants a genuine feel for the racing line — they experience it behind the wheel and not from the passenger seat. With merged data from an exact digital map, along with GPS and video data, the Track Trainer can autonomously guide a vehicle along the racing line. The system is already in its second generation.

The constant comparison of GPS and video data with the digital maps and internal vehicle data was used to guide a vehicle around the north loop of the Nürburgring on 21 October, 2009. Then on 25 May, 2011, the BMW Track Trainer performed a similar feat at the Laguna Seca Raceway in California.

‘The main difference between the sessions at the rack track and trials on the motorway is that we are not alone when driving along a public motorway. That is why we were interested in learning about the developments in the Emergency Stop Assistant Project to ensure safety in our own undertaking,’ says Nico Kämpchen.

The emergency stop assistant is another milestone in the development of automated functions for assisted driving. The system is part of the Smart Senior Initiative of the Federal Ministry of Education and Research in Germany, launched in May 2009. The system incorporates innovative technologies for controlling and pinpointing the location of a motor vehicle and analysing the vehicle environment for added safety. If a driver loses control of a vehicle — for example, due to a health emergency — the emergency stop assistant can detect the situation and autonomously take control of the car and bring it to a safe stop. The system activates the emergency flashers, carefully monitors traffic and guides the vehicle to the side of the road. Then an emergency call is automatically sent out to inform authorities of the situation, including information for emergency medical teams to ensure a quick and efficient response. This is all based on the Advanced Emergency Call function of BMW Connected Drive that is already available as a feature for production cars. This project served as the foundation for the environment recognition function that is used in automated motorway driving.

Research will continue on automated vehicles equipped with advanced driver assistance systems in order to develop assistance functions for the future. Examples are the Parking Assistant and Traffic Jam Assistant in the BMW i3 Concept. Since the BMW i3 Concept is primarily designed as an urban vehicle, it is equipped with functions that make parking easier and driving in congested traffic less stressful. The park assistant, a Bosch product seen elsewhere, automatically parks the vehicle without any driver intervention. The car accelerates and brakes on its own, and shifts gears from forward to reverse as needed when difficult parking manoeuvres are required. The Traffic Jam Assistant helps drivers in monotonous traffic situations and congested areas, taking over so that the vehicle can ‘go with the flow’ and the driver can relax. It maintains a safe distance between vehicles and automatically controls the speed and steering, and is able to stop the car if necessary. As long as the driver keeps one hand on the steering wheel, the vehicle can provide assistance in keeping the car precisely in its lane at speeds up to 40km/h.

Automated driving on motorways provides important information and experience that is essential in developing technologies that assist in keeping vehicles safely on track, and this information will influence future driver assistance systems.

‘The next thing we want to "teach" our prototype is how to deal with road construction sites and motorway junctions. Construction sites are a big challenge because they take all kinds of forms, which makes detection, localisation and determining the right vehicle response quite difficult.’ Nico Kämpchen and his team still have plenty of work to do.


Overview: Volkswagen’s automated driving research: HAVEit

At the final presentation of the E.U. research project HAVEit — Highly Automated Vehicles for Intelligent Transport — Prof. Dr. Jürgen Leohold, Executive Director Volkswagen Group Research, presented his audience with a very unusual Passat estate.

The car is fitted with what Volkswagen is calling Temporary Auto-Pilot. Monitored by the driver, the car can drive semi-automatically up to a speed of 130km/h on motorways. Volkswagen sees the system it uses as a link between today’s assistance systems and fully automatic driving. The Temporary Auto Pilot (TAP) bundles semi-automatic functions — functions monitored by the driver — with other driver-assistance systems, such as adaptive cruise control and the Lane Assist lane-keeping system.

‘Above all, what we have achieved today is an important milestone on the path towards accident-free car driving,’ announced Leohold at the presentation, hosted in the Swedish city of Borås. ‘Nonetheless, the driver always retains driving responsibility and is always in control,’ continued Leohold. ‘The driver can override or deactivate the system at any time and must continually monitor it.’

‘Driving’ an automated Passat.

TAP always offers the driver varies its degree of automation according to the driving conditions, its ability to assess the surroundings, and the states of the system and and (interestingly) the driver. It is intended to prevent accidents caused by driving errors that result from inattention or distraction.

In the semi-automatic driving mode — referred to as Pilot Mode — TAP maintains a safe distance to the vehicle ahead, drives if possible at a speed selected by the driver, reduces this speed as necessary before a bend, and maintains the vehicle’s central position with respect to lane markers. The system also observes posted overtaking rules and speed limits. Stop and start driving manoeuvres in traffic jams are also automated. Drivers must still continually focus their attention on the road, so that they can intervene in safety-critical situations at any time.

In contrast to previous research vehicles such as Junior and Stanley, TAP is based on a relatively production-like sensor platform, consisting of production-level radar-, camera- and ultrasonic-based sensors; these are supplemented by a laser scanner and an electronic horizon.

About HAVEit

The E.U.-funded R&D project HAVEit was set up to develop research concepts and technologies for highly automated driving. The intention is to reduce the driver’s workload, prevent accidents, reduce environmental impact and make traffic safer. Launched in February 2008, 17 European partners from the automotive and supply sector and the scientific community collaborated on the project. Total investments in HAVEit amounted to €28m, of which €17m came from E.U. grants and €11m was contributed by the 17 project partners; of this, €7m was invested by the automobile industry.

The HAVEit consortium consists of vehicle manufacturers, automotive suppliers and scientific institutes from Germany, Sweden, France, Austria, Switzerland, Greece and Hungary. To wit, Continental, Volvo Technology AB, Volkswagen AG, EFKON AG, Sick AG, Haldex Brake Products AB, Knowllence, Explinovo GmbH, German Aerospace Center (DLR), Ecole Polytechnique Fédérale de Lausanne (EPFL), University of Athens, Institute of Communications and Computer Systems (ICCS), University of Applied Sciences Amberg-Weiden, Budapest University of Technology and Economics, Universität Stuttgart, Institut für Luftfahrtsysteme, Würzburg Institute of Traffic Sciences GmbH, Institut National de Recherche en Informatique et en Automatique (Inria), Institut Français des Sciences et Technologies des Transports, de l’Aménagement et des Réseaux (IFSTTAR).

Text and design copyright © Under the Skin 2010-2011. We recommend Firefox.