Den här sidan är utskriven från Högskolan i Halmstads webbplats (www.hh.se). Texten uppdaterades senast den 2017-09-26. Besök webbplatsen om du vill vara säker på att läsa den senaste versionen.
We are witnessing major technological and societal developments in personal and commercial mobility. This includes ever smarter driver assistance systems, vehicles that achieve efficient and reliable operation by continuously monitoring and understanding sensor data, as well as rapid advances towards autonomous cars, both on the technological and legal front. Many innovations are based on the increasing amount and variety of data exchanged between vehicles and with the infrastructure.
Development of intelligent vehicles combines many disciplines, including data mining, artificial intelligence, robotics, signal analysis, perception, machine learning, big data, human factors, and interaction design. Our cutting edge research combined with the capabilities and assets of our partners allows CAISR to push the boundaries of the feasible and economically viable, creating scientifically valid and practical solutions.
Our aim is to make significant contributions in the intelligent vehicles area in close collaboration with our partners, both inside the University and outside, academic as well as industrial, nationally as well as internationally. We have a track-record of successful industry cooperation, particularly in Sweden, and are continuously developing collaborations with academic and commercial partners in Europe, Japan, and the USA.
Our intelligent vehicles project portfolio builds on our strengths in data mining on the one hand and autonomous systems on the other. This is reflected in our expertise in data-driven modeling, knowledge representation, self-monitoring, localization, perception, mapping, and motion generation.
Within machine learning, we develop methods and tools to analyse and leverage the wealth of data that can be collected on modern vehicles. The aim of this line of research is to model and detect behavior patterns in fleets of vehicles (self-awareness), and apply these models and methods to improve operational efficiency (situation awareness). We also focus on keeping the human in the loop through joint human-machine learning.
Concerning autonomous systems, our focus is on making next generation self-driving and autonomous vehicles aware. This ranges from interactive semantic mapping as well as multi-vehicle motion planning and plan adaptation (especially in environments shared with humans and under shared control where the safety aspects and interaction on various levels is of highest importance), to understanding and improving the modalities and interfaces that are most appropriate for bringing technological advances to the end user (human awareness).
In 2016 we have finished a project In4Uptime (co-funded by Vinnova, AB Volvo and Recorded Future) on adapting service contracts and maintenance plans to the needs of individual customers and vehicles based on data from different sources, such as on-board, off-board, structured, unstructured, private and public, and by combining information and finding common patterns. It builds on results from earlier ReDi2Service project, where on-board data was analysed to detect unusual vehicle behavior and link these patterns to causes before vehicle failures occur. This work continues today in collaboration with Aftermarket group at Volvo Trucks and Volvo Busses.
Since product quality is one of the top priorities for commercial vehicle manufacturers, a new ARISE project starting this year aims to develop algorithms for early detection of quality issues and their analysis, integrating multiple available data sources. Our solutions will allow for automatic and semi-automatic highlighting and prioritising of the most relevant issues, based on online and incremental algorithms especially suited for early detection of problems in the product launch phases.
We are also analysing driver behaviour, from several different perspectives and with several different goals in mind. Safe and smooth driving is critical in forklifts operating within warehouse environment; high levels of work related stress and disturbed sleep are a dangerous combination contributing towards diseases and poor workplace performance in bus drivers; there is need to identify and quantify, from extensive field operational data, factors that affect fuel consumption.
The goal with SAS2-project is to develop a safety system that better can handle the increasing complexity (and the implied higher requirements on safety system) in the environments where AGV:s operate. The approach is to use 3D perception along with methods for detect, track and identify objects in the environment, such that actions of moving objects can be foreseen and decisions can be made based on object identities (humans and other trucks). Such a situation aware safety system will, ultimately, increase the productivity and efficiency of AGV systems.
Autonomous vehicles and robots are being introduced into public and domestic environments to conduct useful tasks, but for such technologies to perform effectively, it is crucial to ensure safety, trust, and acceptance. Toward this, our AIR project aims to facilitate mutual recognition of actions and intentions between humans and the autonomous systems in homes, industry, and public settings. In collaboration with Skövde University, Örebro University, and RISE Viktoria, work at CAISR is focusing on efficient and robust visual recognition using dense motion features and optical flow, generation of robot motions capable of clearly communicating intentions, and investigating novel modalities for robot recognition and behavior.
An unprecedented growth of data, fed by novel technologies, user behaviour and business models, is one of the most dramatic and important developments in the society today, and Intelligent Vehicles is an important part of that. There is an enormous commercial, societal and environmental potential in exploiting this information. The very fact that data is now in such an abundance makes a qualitative, not only quantitative, difference and opens up the possibility to develop new tools that are vastly superior to today's technologies. In the BIDAF project we have created a strong distributed research environment for Big Data Analytics. The scientific objectives concern advanced, near real-time analytics on uncertain data with high volume and velocity through machine learning techniques; key challenges include development of a computational platform and machine learning algorithms suitable for handling both opportunities and challenges with massive data, as well as to provide analytics methodology and high level functionality to make the value in massive data easier to access. The BIDAF consortium consists of RISE, Skövde University and Halmstad University.
Our intelligent vehicles project portfolio builds on our strengths in data mining on the one hand and autonomous systems on the other.