Conference Day 1 - 26th April 2016

Conference Day 1

Tuesday 26 April – Conference Day 1

  1. Registration and Refreshments

  2. Chair's Opening Remarks

    Salah Hadi | Global R&D Director Vision & Night Vision Systems of Autoliv Vision Systems, Sweden

OEM and T1 vision and insights from Europe, Asia and US

Where do they want the IS industry to focus their efforts in the short, medium, long term?

  1. Driving the Future, Connected Autonomous Vehicle

    Dr Bakhtiar Litkouhi | Technical Fellow and the Manager of Automated Driving and Vehicle Control System of General Motors, USA

  2. HITACHI, a global T1 player for Autonomous Driving

    Dr Massimiliano Lenardi | Laboratory Manager - Senior Researcher of HITACHI, France and Germany

    Hitachi’s Automotive Vision targets the supply of reliable components and telematics solutions enabling self-driving vehicles. Insights will be given on technologies and services needed to deploy autonomous driving, in particular on those provided by Hitachi, from total-sensing to control and communication technologies.

  3. Challenges of High Resolution Automotive-Multi-Camera-Systems

    Dr. Vikram Narayan, Algorithm Developer, Panasonic Automotive & Industrial Systems Europe GmbH

    Panasonic Automotive & Industrial Systems is a global Tier 1 supplier supporting automotive market with camera systems for various driver assistance systems for imaging, recognition, classification etc.
    Safe and reliable autonomous vehicle movements require multiple sensor support. A single cameras can be seen as a passive sensor using the visual spectrum comprising of millions of independent sensor elements.
    Multiple cameras, which support an application brings high demands concerning the:

    • Sensor limiting Factors
    • Data Avalanche
    • Efficient Data Reduction and Information Extraction
  4. Panel-IS Systems, Applications and Timelines: What Are The Key Areas for the Short and Medium Term?

    Panellists include: General Motors Research & Development and Hitachi Europe Ltd

    • How are customer expectations changing for IS applications in cars?
    • What are the latest challenges of integrating all the modules?
    • What are the key challenges OEMs want the IS ecosystem to address?
    • How should components suppliers approach the auto market – what do they need to consider to industrialise their ideas?
    • What is the future of interoperability?

    Salah Hadi, Global R&D Director Vision Systems, Autoliv, Sweden

    Panellists include:

    Dr Bakhtiar Litkouhi, ‎Technical Fellow. Manager: Automated Driving & Vehicle Control System, General Motors Research & Development, USA

    Dr Massimiliano Lenardi, Laboratory Director, HITACHI Europe Ltd, France and Germany

  5. Networking Refreshment Break

  6. Euro NCAP’s Vision Towards 2020

    Aled Williams | Programme Manager of Euro NCAP, UK

    The presentation will show the current developments of test and assessment protocols and will outline Euro NCAP’s vision towards 2020 and beyond.

  7. Programmable Automotive Headlights: High-Performance Lighting and Imaging at 100 Miles Per Hour

    Srinivasa Narasimhan | Associate Professor Robotics Institute of Carnegie Mellon University, USA

    • Gain insign into a new design for a headlight that can be programmed to perform several tasks simultaneously and that can sense, react and adapt quickly to any environment with the goal of increasing safety for all drivers on the road
    • Explore the engineering challenges in building this headlight as a high-throughput, low-latency platform for computational imaging and lighting
    • Assess experiences with the prototypes developed over the past two years
  8. Networking Lunch for Speakers and Delegates

Day 1 Stream A

Stream A: Latest Sensor and Lens Technologies

Driving improvement in performance, power consumption, quality, robustness and efficiency to solve today’s most pressing needs.


Chair:  Salah Hadi  |  Global R&D Director Vision & Night Vision Systems of Autoliv Vision Systems, Sweden

  1. The Automotive Lens - Characteristics and Future Trends

    Winwe Qiu | General Manager of Ningbo Sunny Automotive Optech Co., Ltd, China

  2. Exploring Global Shuttered CMOS Image Sensors

    Mario Heid | General Manager OVT Europe of OmniVision, Germany

  3. BrightEye™ - Eyes for ADAS

    Dr Ofer David | CEO of BrightWay Vision, Israel

    Gated imaging has shown great potential in automotive imaging. Raw video imagery by day and night for obstacle detection functionalities will be presented, along with a comparison of gated and passive in harsh weather conditions. Simplified traffic sign localization in 3D will also be discussed. 

    Take this opportunity to see a demo of this exciting technology.

  4. Networking Refreshment Break

  5. ‘facetVISION’ - Boldly Go Where No Camera System Has Gone Before

    Andreas Brückner | Head of Microoptical Imaging Systems Group of Fraunhofer IOF, Germany

    There is a constant trend for miniaturization of digital cameras which is mainly driven by mobile devices like smartphones but also extremely valuable for applications in automobile and machine vision. It pushes the shrinking of opto-electronic, electronic and optical components. While opto- and micro-electronics have made tremendous progress, the miniaturization of optics still struggles to keep up. The demands for higher image resolution and large aperture of the lens (both driven by smaller pixel size) conflict with the need for a short focal length and a simple, compact design in terms of miniaturization. Array cameras inspired by the smallest known vision systems in Nature – the compound eyes – offer a way out of the dilemma.

    The contribution provides an illustration of the fundamental limits of the miniaturization of digital imaging systems. It is shown that these limits can be at least partly overcome by the convergence of microelectronics, microoptics and image processing applied in array cameras. The basics about the wafer-level optics fabrication technology are presented in order to demonstrate its potential for high-precision, parallelized production and thus for reducing the production costs. Finally, the contribution gives examples of realized demonstrators for array imaging sensors and cameras of smallest size that are able to overcome the scaling limits of traditional optics and thus to go where no camera has gone before.

  6. The Future of Stereo Cameras in Automobiles

    Dr Michael Chiu | Chief Technology Officer of Automation Engineering Incorporated, USA

    • Critical attributes of stereo cameras and evolution into the future
    • Enabling processes and components for stereo cameras, including image sensors, manufacturing & test processes
  7. Wrap-Up Discussion and Closing Remarks

  8. Networking Drinks Reception and Vehicle Demonstration

Day 1 Stream B

Stream B: Image Processing & Computer Vision

Chair: Dr. Hamma Tadjine, Senior Project Manager, IAV

  1. Heterogeneous Computing for Real-Time Computer Vision

    Kari Pulli | Senior Principal Engineer, Imaging and Camera Technologies Group of Intel, USA

    • Specialized accelerators can be orders of magnitude more efficient than general-purpose hardware, but they can be difficult to program
    • Recent standards such as OpenCL and OpenVX can make such accelerators easier to access
  2. Deep learning - Tailoring Neural Networks for Automotive Applications

    Etienne Perot | Vision Research Engineer of Valeo, France

  3. Smart Imaging

    Mayank Mangla | ADAS Imaging Architect of Texas Instruments, USA

    Image processing industry has been inundated with choices, ranging from fully programmable to FPGA based architectures. While programmability provides flexibility it lacks in performance and cost competitiveness. FPGAs are very efficient at what they are designed to do but fail miserably when requirements change even slightly.

    In this presentation we demonstrate highly efficient yet programmable ISP architecture from Texas Instruments TDAxx SoCs. The architecture achieves full pixel processing in HW but takes help of a lightweight control processor for decision making. Through this architecture TDAxx devices have been able to target a much wider range of applications than what the devices were designed for.

    The concept is further explained with the help of following case studies

    1.GPU less 3D Surround View
    2.Camera Monitoring Systems with support for LED Flicker Mitigation
    3.Obstruction Detection
    4.Sony Sensor Digital Lateral Overlap   

  4. Power-Efficient Implementation of Deep Neural Networks for Autonomous Driving Systems

    Jeff VanWashenova | Director of Automotive Market Segment of CEVA Inc., USA

    Many automotive companies are developing deep learning capability for active safety and autonomous driving.  One challenge many companies are facing is moving from CPU/GPU development environments into a computing platform that is more optimized for high volume applications.

    CEVA has both the silicon IP that is cost effective and efficient, but also a means to migrate fully trained networks to this optimized vision core.  This discussion will focus on the phases of deep learning from R&D to production, the benefits of platforms for each phase and a migration path to production system.

  5. Networking Refreshment Break

Stream B: Driver Monitoring Systems

  1. Gesture Control and Drive Monitoring: The Challenge To Make Innovative Image Processing Technology Automotive-Grade

    Sascha Klement | Managing Director, CTO of gestigon GmbH, Germany

    • Features and use cases for gesture control and driver monitoring
    • Technical requirements in automotive applications
    • KPIs for assessing the quality of such systems
    • Testing and quality assurance strategies 
    • Social challenges in the transition from innovative prototypes to automotive-grade series production
  2. ToF Based Driver Monitoring: A Self-Monitoring Approach Based on Mass Market Components

    Dr. Bernd Buxbaum | CEO/CTO, Founder of pmdtechnologies, Germany

    On semi-autonomous and autonomous driving, the driver awareness remains an important safety factor. For safe driver monitoring, ToF (Time-of-Flight) provides unique robustness features on a reasonable cost target.  

    • Integrated availability recognition: intrinsic amplitude value with each distance including saturation recognition
    • Real monitoring of whole sensor chain: reference channel in each measurement cycle
    • Long-term availability: robust availability, even on mechanical stress
    • Usage of mass market components only – all system components are proven in use
    • Moderate calculation effort for accurate distance information
    • One camera head for eye lid recognition and distance measurement


  3. Developing Driver Monitoring System Algorithms for Tomorrow's Vehicle

    Alexandru Drimbarean | Vice President Advanced Research of FotoNation Ireland

    Driver and passenger safety is one of the main concern for car makers. Driver’s undivided attention to the traffic is essential to avoid any serious accidents. The National Highway Traffic Safety Administration conservatively estimates that 100,000 police-reported crashes are the direct result of driver fatigue each year. This results in an estimated 1,550 deaths, 71,000 injuries, and $12.5 billion in monetary losses. These figures may be the tip of the iceberg, since it is currently difficult to attribute crashes to sleepiness. Also, distracted driving such as using a cell phone, texting and eating is the cause of 1 out of 5 crashes in the US. In 2012, 3,328 people were killed in crashes involving a distracted driver. Finally, the race toward semi-autonomous and fully-autonomous driving is accelerating the need to have a driver monitoring system is paramount since the vehicle needs to know what the driver state is before it gives back control.

    The presentation will begin by detailing the key aspects related to the image acquisition, processing and computer vision technologies with emphasis placed on eye and gaze detection as well as head location and orientation calculation. These are required to continually monitor and assess the driver state in order to detect conditions such as excessive drowsiness or lack of road attentiveness that could potentially lead to accidents.  Then a more advanced use case, namely Driver identification implemented using face recognition technology is introduced. This is becoming increasingly relevant for car personalization: seat adjustments, mirror adjustments mirror settings as cars are being used by multiple persons as well as for security particularly important for theft prevention. Finally we address implementation specifics, emphasizing the flexibility that software can offer in terms of where the camera could be located and how hardware acceleration could bring performance and thermal management advantages.

  4. Wrap-Up Discussion and Closing Remarks

  5. Networking Drinks Reception and Vehicle Demonstration