Hahn M.,Daimler Group Research |
Barrois B.,Daimler Group Research |
Kruger L.,Daimler Group Research |
Wohler C.,Otto Group |
And 2 more authors.
3D Research | Year: 2010
This study introduces an approach to model-based 3D pose estimation and instantaneous motion analysis of the human hand-forearm limb in the application context of safe human-robot interaction. 3D pose estimation is performed using two approaches: The Multiocular Contracting Curve Density (MOCCD) algorithm is a top-down technique based on pixel statistics around a contour model projected into the images from several cameras. The Iterative Closest Point (ICP) algorithm is a bottom-up approach which uses a motion-attributed 3D point cloud to estimate the object pose. Due to their orthogonal properties, a fusion of these algorithms is shown to be favorable. The fusion is performed by a weighted combination of the extracted pose parameters in an iterative manner. The analysis of object motion is based on the pose estimation result and the motion-attributed 3D points belonging to the hand-forearm limb using an extended constraint-line approach which does not rely on any temporal filtering. A further refinement is obtained using the Shape Flow algorithm, a temporal extension of the MOCCD approach, which estimates the temporal pose derivative based on the current and the two preceding images, corresponding to temporal filtering with a short response time of two or at most three frames. Combining the results of the two motion estimation stages provides information about the instantaneous motion properties of the object. Experimental investigations are performed on real-world image sequences displaying several test persons performing different working actions typically occurring in an industrial production scenario. In all example scenes, the background is cluttered, and the test persons wear various kinds of clothes. For evaluation, independently obtained ground truth data are used. © 2010 3D Display Research Center and Springer-Verlag Berlin Heidelberg.
News Article | December 6, 2016
Stuttgart/Berlin, 06-Dec-2016 — /EuropaWire/ — How is artificial intelligence (AI) influencing tomorrow’s mobility? How can we use its ideas today? How intelligent will the car of the future be? And above all: what role will people play in this? These were the central questions discussed by Mercedes-Benz experts at the fourth Future Talk in Berlin in dialogue with scientists, engineers and journalists. In the past few years the Mercedes-Benz Future Talk has dealt with the subjects of utopia, robotics and virtuality. This time, the focus was on the integration of artificial intelligence in the field of mobility and the interaction of humans and machines. Already in the 1960s researchers expected a major breakthrough in the development and application of artificial intelligence, but the human world still proved too complex for digital computers. However, due to the triumph of the internet, the mass of data that has become available with this and the huge increase in computing power of today’s computers, artificial intelligence is now entering people’s lives and also offers big opportunities and potential for the future of the automobile. “Artificial intelligence is a key future topic for Mercedes-Benz, in-car and beyond, such as in the fields of mobility services or in development and production”, says Anke Kleinschmit, Head of Daimler Group Research. “Artificial intelligence has ceased to be science fiction and the progress in autonomous driving is an impressive proof of this. Likewise, AI already assists the development phase and production by providing intuitive access to global knowledge and knowhow – Always tailored to the individual needs, experiences and knowledge of the employee. In addition to the technical development and data security, a basic prerequisite for the sustained success of artificial intelligence in all application cases is the acceptance by society and consumers. “Artificial intelligence will only be successful on a long-term basis if we succeed in building up trust between man and machine”, says futurologist Alexander Mankowsky. “We must define the division of tasks between human and artificial intelligence.” A necessary prerequisite is also to be aware of what artificial intelligence is able to do and what it isn’t. Because ultimately it always needs human participation and is based on human development. But it can support to make and examine decisions and therefore reach optimal results in a shorter space of time. The philosophy of Mercedes-Benz is – always to put human being at the centre of all activities. Cognitive vehicles: the car as a control centre for individualised AI An important objective of Mercedes-Benz’ activities relating to artificial intelligence is the development of cognitive vehicles. They are not only able to respond to certain situations; they even have enough knowledge about their environment to be able to act autonomously on this basis. Coupled with corresponding services they could become the fundament for a holistic mobility eco-system of the future. For example, they could autonomously analyse the current traffic situation for all forms of transport and draw up an individual mobility plan that suits the customer’s personal daily routine and mood. In addition, household robots and delivery drones could be linked to the system with the cognitive car as the control centre for this. Unlike smartphones and wearables, the car would surround the person and become a surrounding for a digital experience. It could analyse the driver’s behaviour, interpret needs and adapt accordingly. It would be able to identify what he or she wants in certain situations and what he or she needs. Examples of this are playing the right music to suit the current mood, setting the most pleasant temperature or developing services relating to health and safety. Moreover, the cognitive vehicle would offer self-determined access to an individualised artificial intelligence which supports human beings, entertains them and could even challenge them intellectually. Image and pattern recognition as an important milestone on the way to autonomous driving To successfully embark on this path, vehicles must be able to acquire knowledge about their environment as well as analyse it. This machine learning already plays an important role for autonomous driving as of today. Mercedes-Benz is working intensively on the further optimisation of automatic image and pattern recognition for driver assistance systems and autonomously driving vehicles. A decisive topic here is the interaction of cameras, sensors and the associated computing units. The system breaks down the pictures of road scenes into abstract segments with coloured marking. In this way it identifies buildings, vehicles, persons, trees and pavements among other things and reliably finds traffic lights as well as smaller dangerous obstructions on the road. Based on this, the autonomous vehicle analyses the traffic situation, predicts the behaviour of other road users and decides on its own behaviour. “In daylight many systems for image and pattern recognition, on the market are reliable”, says Dr Uwe Franke, responsible for image recognition/signal processing and sensor fusion in the Mercedes-Benz development department. “Meanwhile, our system even offers top level results at night and that is a major development. The next step is about recognising and interpreting people’s gestures and facial expressions.” It is the recognition of gestures, facial expressions and people’s understanding of machine behaviour that makes an operative interaction between man and autonomous vehicles possible at all. On this basis trust can be created between humans and machines. Vehicles must be able to make it clear that they recognise pedestrians and pay attention to them. Pedestrians must receive information about where an autonomous vehicle is going, how it will behave in the next few moments and how they should behave themselves. Finding and making faster use of ideas and potential with AI Mercedes-Benz is not just using artificial intelligence with regard to its vehicles. Among other things the company is testing self-learning systems in the observation of technology trends, in the interpretation of development and test data as well as for the industrial maintenance of its production and manufacturing facilities. Artificial intelligence can make a decisive contribution to diagnosing technical problems. For example, until now production maintenance staff either had to search through huge amounts of documents or fall back on their personal experience to get information about machine defects. The tested system handles documentation with natural language processing and serves as a semantic search engine. Unlike a keyword-based search engine the focus is on the meaning of the request. This enables requests in the form of various fault descriptions, for example “oil is leaking” or “leaking pipes”. In this way repair and maintenance processes can be speeded up and made more efficient. Mercedes-Benz to work with a leading AI institute in future Mercedes-Benz works with numerous renowned research institutions in the field of artificial intelligence. For years Daimler has been expanding its network to universities with innovative instruments such as “Forschungscampus”, tech centres, shared professorships and industry fellowships and start-up incubators and accelerators. In addition, the company will become a new member of the “MIT CSAIL Alliance Program” in the near future. With 50 research groups and around 1000 members of staff, the Computer Science and Artificial Intelligence Laboratory (CSAIL) of the Massachusetts Institute of Technology (MIT) is one of the leading institutes worldwide in the field of IT and AI. “Mercedes-Benz is rigorously advancing its research and development in different directions so as to continue to play a pioneering role in the automotive industry in the development and application of artificial intelligence. The new cooperation with the MIT ideally complements this. The partnership enables us to benefit even more directly from the research results of a leading world institute and to network with the best brains”, Anke Kleinschmit emphasises. “We aim to continue to play a leading role in shaping the future of mobility – with new mobility concepts, with cognitive cars and services which focus on people and make their daily lives easier and better.” As well as Anke Kleinschmit, Head of Daimler Group Research, the Daimler employees participating in this year’s round table discussion were futurologist Alexander Mankowsky, Dr Uwe Franke, responsible for image recognition/signal processing and sensor fusion and Patrick Klingler from IT Innovation Management. The external experts included Dr Miguel Nicolelis, Duke Center for Neuroengineering and Prof. Jürgen Schmidhuber, IDISIA Swiss Research Institute for Artificial Intelligence. Future Talk participant Prof. Jürgen Schmidhuber sees numerous fields of application for artificial intelligence in future. “It will take decades at most rather than centuries for us to develop true artificial intelligence. Artificial intelligences will learn almost everything that people can do – and much more besides. Their possibilities for shaping the future, are really only limited by our imagination.” Also Dr Miguel Nicolelis is convinced: “Artificial intelligence can lead to some sort of machine intelligence. This can be helpful in the future if it is under the control of human intelligence, not the other way around.” The Future Talk is a dialogue format successfully established by Mercedes-Benz in 2013. By exchanging ideas with vanguards from various disciplines, the brand shares its visions and, as the inventor of the automobile, demonstrates its expertise in shaping a desirable, mobile future. The focus topics to date are representative of the variety of this meta topic. This year’s Future Talk continues the discussion of the last few years on the subjects of “Utopias” (2013), “Robotics” (2014) and “Virtuality” (2015).
Jander K.,University of Hamburg |
Braubach L.,University of Hamburg |
Pokahr A.,University of Hamburg |
Lamersdorf W.,University of Hamburg |
Wack K.-J.,Daimler Group Research
International Journal on Artificial Intelligence Tools | Year: 2011
Business process management is a challenging task that requires business processes being described, executed, monitored and continuously enhanced. This process management lifecycle requires business as well as IT people working together, whereby the view on business process is quite different on both sides. One important means for bridging the gap between both consists in having a modeling notation that can be easily understood but also has a precise semantics and can be used as a basis for workflow execution. Although existing approaches like BPMN and EPCs aim at being such as notation they are already very activity oriented and do not consider the underlying motivations of processes. Introducing the goal oriented process modeling notation (GPMN) a new language is presented that has the objective of bringing together both sides by establishing higher-level modeling concepts for workflows. This results in an increased intelligibility of workflow descriptions for business people and greater consideration for the way processes are described on the business side. The core idea of the approach consists in introducing different kinds of goals and goal relationships in addition to the established activity-centered behavior model. The applicability of the approach is further illustrated with an example workflow from Daimler AG. © 2011 World Scientific Publishing Company.
Braubach L.,University of Hamburg |
Pokahr A.,University of Hamburg |
Jander K.,University of Hamburg |
Lamersdorf W.,University of Hamburg |
Burmeister B.,Daimler Group Research
Studies in Computational Intelligence | Year: 2010
Many companies consider business process management strategies a fundamental source for successful business operation. Despite this importance of business processes a conceptual and operational gap still exists between the business and the IT view of processes. In this paper we argue that an important reason for this gap is the strong focus of IT on the behaviour and execution perspective of workflows while more abstract and higher-level process properties are often neglected. This is especially apparent in the way processes are modelled and described on the IT-side using state of the art modelling approaches like BPMN. The presented Go4Flex research project, which is conducted in cooperation with Daimler AG, has the objective of bringing together both sides by establishing higher-level modelling concepts for workflows, which results both in increased intelligibility of workflow descriptions for business people and greater consideration for the way processes are described on the business side. The core idea of the approach is to strengthen the context perspective of a workflow by introducing different kinds of goals and goal relationships in addition to the established activity-centred behaviour model. The applicability of the approach is further illustrated with an example workflow from Daimler AG. © 2010 Springer-Verlag Berlin Heidelberg.