Published on March 13, 2018
A virtual driving instructor for megacities
Text | Einleitung
Schema Blabla
Particularly in densely populated cities and regions, driving schools are faced with the challenge of catering for a steadily growing number of learner drivers. “In China, this demand is met with prescribed training lessons in driving simulators,” explains Polina Häfner of the KIT Institute for Information Management in Engineering (IMI). “Only those who have had the minimum required number of driving lessons with certain driving exercises in the simulator are awarded a driving licence.” However, a virtual driving lesson only makes sense if learner drivers experience the simulation in conditions that are as realistic as possible.
In the “Smart Immersive Environments” research team at the IMI, the scientists and students headed by Institute Director Professor Jivka Ovtcharova deal with precisely such intelligent, virtual environments. “We are seeking to achieve diving into a non-existent environment by deceiving human senses with technical apparatus along the lines of Augmented and Mixed Reality. We are not conducting classical pure research but are focusing very strongly on industry. Over the last few years, we have developed the knowhow, a high-performance infrastructure and state-of-the-art technology that we require for this purpose at the Institute,” explains Professor Ovtcharova. Initial simulated driving activities already commenced at the Institute in 2010. A sort of box to sit in, fitted with a car seat, a steering wheel and pedals, was turned into a driving simulator. By a fortunate coincidence, the research team received a real small car in order to professionalise the driving simulator. “Although it was difficult to transport the car in the Institute’s facilities without an engine, the effort was worthwhile,” scientist Häfner stresses. “Working with real products is fun and provides motivation.”
China as a marketing opportunity
A meeting with a KIT alumnus proved to be another key event in the development of the driving simulator. The qualified engineer with Chinese roots discovered the market potential at KIT during a trip with a delegation in 2016. And he simultaneously supplied a concrete business concept with a market volume of more than 100 billion for driving lessons in China – as a virtual driving instructor at Chinese driving schools. “First of all, I thought this was utopian. In Germany, this kind of thing would probably not be authorised,” Professor Ovtcharova says. “But China’s market and its framework conditions can turn it into an option.” Chinese investors and a business start-up paved the way for the commercial use of the driving simulator developed at KIT.
In spite of initial language barriers, things swiftly progressed from the basic technology through initial contact to cooperation in the framework of the technology transfer project “DriveSim” (Driving Simulator), in which the KIT scientists are cooperating with the Chinese development partners TuoBaBa Technology (TBB) and the Jiangyin Sino-German Technology Transfer Center as a further cooperation partners in the implementation of the virtual trainer. Here, KIT was in charge of hardware configuration and software solutions, while the Chinese partners prepared the local server infrastructure, remote maintenance and applications for mobile terminal apparatus.
Top-standard driving simulation
“There already are a number of driving simulation solutions,” project coordinator Häfner explains. “So far, however, many solutions have been based on permanently programmed or artificially created worlds.” Here, routes are programmed in advance, for example according to what a designer has proposed. In contrast, DriveSim is based on real data from geo-information systems (GIS). During driving in a simulator, the algorithm that has been developed builds up the environment almost in real time as a close-to-nature, virtual world – no fantasia lands but real areas, buildings and street situations. “We have also opted for a familiar human-machine interface in a real car,” Häfner stresses. “Here, learner drivers are to gain an optimally realistic impression of driving by directly and realistically feeling forces and controls such as the steering wheel, pedals or gears.”
In order to achieve such an optimally realistic feeling of driving, a complete platform with software and hardware as well as corresponding interfaces with the vehicle were put into practice. Interventions in the vehicle are accomplished via the so-called CAN-Bus (Controller Area Network), a control apparatus inside the vehicle that links sensors and actors and regulates data transfer between the components. DriveSim uses this network to control the non-motorised vehicle, incorporating the displays in the cockpit. “Every manipulation in the vehicle, such as steering, shifting gears, accelerating or braking, is reflected in the virtual world while simultaneously being made perceptible for the learner driver with the aid of a force-feedback method,” Häfner explains.
The vehicle is surrounded by projections or monitors to provide a view of the three-dimensional world, virtual reality. In connection with the virtual reality software that has been developed, this enables changing driving environments to be generated and represented. An intuitive authoring tool in the shape of an app is also part of the project. In future, it will enable driving instructors to swiftly and comfortably configure the traffic exercises tailored to an individual learner driver. In order to ensure that processing can performed in real time during a journey and that it is possible to directly evaluate driving behaviour, the entire platform and collection of data – ranging from static GIS data to dynamic driving behaviour – are based on a semantic data model. “The technical environment has been enhanced and validated in the course of the project. Now two vehicles have been retrofitted as simulators: the small car at the Institute and, in China, a midsize car typical of those that are employed at Chinese driving schools. Simulating unevenness in road surfaces and giving the learner driver a feeling of accelerating will be further major milestones in the project,” Professor Ovtcharova, whole holds doctoral degrees in mechanical engineering and informatics, reports.
Learning driving virtually
In spite of the virtual driving lesson, the KIT driving simulator allows intuitive manipulating in a real motorised vehicle. When looking out of the window or at one of the car’s mirrors, one will discover a close-to-nature, three-dimensional world that, if required, can be a representation of familiar networks of roads in one’s neighbourhood. The learner drivers are provided with feedback and learning support in the form of a special training app. “It is possible to track driving behaviour. The app records progress that the learner driver has attained,” Häfner explains. “If needed, supporting information on driving is provided in the app or in the virtual world.” The training results are also available for the driving instructor. In order to achieve gradual progress in learning, individual training exercises can be compiled with varying complexity regarding traffic density, weather conditions and street configurations. Currently, DriveSim is being developed for local use in driving schools. If the driving simulator enjoys acceptance, the long-term goal will then be to offer several centres in urban districts for “in-between” driving lessons.
Full speed ahead
“In addition to China’s demand, inquiries are coming in from driving schools in Germany that are interested in the project. For example, we have contact with a driving school in Karlsruhe, which currently training around 5% Chinese learner drivers. Here, we can benefit from the driving instructor’s specialised knowledge,” Häfner reports. The virtual driving lesson can certainly also provide added value for German driving schools by focusing on exercising individual driving tasks. This is above all an advantage for very young learner drivers, but also for older drivers wishing to get fit for the road again after not having driven for a longer period. Professor Ovtcharova is optimistic about the future. “There is a considerable range of options for the further use of our development,” she explains. “Our algorithms and software would also represent added value in the field of autonomous driving. Especially with more presence in the media we are hoping to see progress being made with the next major topic and new opportunities opening up.”
comments about this article
No comments