Day 2 | 2017年 10月 6日（金）
Sandeep Sovani (ANSYS)
Mauro Barbieri (Ferrari)
Dave Hartfelder (General Motors)
Director of System Safety
Ever growing automotive embedded control system complexity directly translates into greater potential for configuration management issues and opportunity to miss the implications of system and software iterations. This is particularly challenging in the functional safety case. Utilizing a specialized functional safety analysis and process execution tool can provide configuration management, efficiency and quality benefits allowing System Safety engineers to focus their expert skills on developing and analyzing robust, risk minimized safety critical systems. Following a model based systems engineering approach can further expand these benefits. This presentation will explore the benefits of following a model based systems engineering approach along with the implementation of a functional safety analysis and process execution tool to minimize the opportunity for systematic issues in engineering safety critical systems.
Alberto Bassanese (Lucid Motors)
Manager of Multiphysics and Optimization
Michael Wagner (Edge Case Research)
We need more than a system-level "test-fail-patch-test" cycle to deploy safe autonomous vehicles at scale. Traditionally, we turn to software-safety standards such as ISO-2626 that are based on carefully tracing each type of testing to a corresponding design or safety requirement. However, it is now widely recognized that existing standards face challenges when applied to autonomy. Edge Case Research and Carnegie Mellon have defined exactly how autonomy challenges traditional software-safety standards, and have identified promising solutions. These include the use of powerful simulation engines paired with artificial intelligence for exploratory robustness testing that efficiently finds "black swans" in autonomy software.
株式会社本田技術研究所 第5技術開発室 第1ブロック 主任研究員
車両開発フローはシミュレーションによる設計とバーチャル検証に進化しつつあり、MBD 等の開発アーキテクチャによって、全体最適な設計が可能となってきた。一方、EMC 設計は、ノイズ源・伝搬経路・被害機器を開発する部門間で相互影響を把握する手段が無いため最適化が中々進まない。この課題を解決するため、各EMC 要素が車両全体に及ぼす影響を、共通認識可能な縮退モデル化を試みたので、その事例を紹介する。
慶應義塾大学大学院 システムデザイン・マネジメント研究科 教授・工学博士
IoT (Internet of Things) で形づくるIndustrial Internet Systems (IISs)は，いわゆるSystem of Systemsを形成し，その複雑度合いは極めて高くなる．こうしたシステムを全体として安全で耐久力のあるものとするには，アーキテクチャに基づいてマネジメントを続けることが求められる．また，このためには，システム解析を行える体制を常に維持することが必要となる．MBSE（モデルに基づくシステムズエンジニアリング）はこうした問題に対処するための一つの方向性を見いだせるものと期待されている．自動運転システム（レベル3）を搭載する自動車を取り巻く交通システムを事例に，その考え方を紹介する．
日本大学生産工学部 自動車工学リサーチ・センター 教授・上席研究戦略アドバイザー
Optimization of multiple physics problems for design verification process in Kyungshin
Byeong Woo Kim (Kyungshin)
Senior Manager, CAE Section
The computer based design verification has been applied mainly in single physics simulation. As computer technology and commercial software grows faster, more realistic and comprehensive engineering simulations are feasible in manufacturing industries. For example, the shape optimization for combined external loads under cyclic heats and dynamic forces had been developed. The simulation process can predict the fatigue life for design specifications before physical testings. Kyungshin has plans to expand multiphysics simulation to the system level simulation and engineering knowledge management system. This presentation introduces overall simulation efforts and further topics in developing automotive components in Kyungshin Corp.
Leading New ICT
The Road to Automotive Digital Transformation
Francis Lam (Huawei Technologies Co. Ltd.)
Director, IT Product Management
Olaf Kath (ANSYS)
More than 60 percent of a new vehicle’s cost comes from its advanced electronics and software systems. Since many of the human-machine interface (HMI) functions guided by these electronic systems are mission-critical, it’s essential that all automotive systems work together with complete reliability. That means tens of millions of lines of software code that control advanced driver assistance systems (ADAS) must be flawless. If your embedded software and HMI development process relies on generic database tools and time-consuming methods to manually generate and verify control code, any human error can be costly. On the development end, you’re likely to not receive feedback until the code is compiled and run on an expensive test vehicle. Afterwards, you’ll likely need to make extensive changes to the embedded code and HMI following each driving scenario, and this can be a long and expensive process.
Learn how to overcome these embedded software and HMI development challenges using the ANSYS model-based approach. You will learn how to develop ISO 26262-qualified and AUTOSAR-compliant systems, providing:
• An estimated 40 percent savings in costs
• Greater control over complex system architecture tasks
• A higher level of system reliability, safety and security
• A reduction in physical testing investments
Robert Myoung (ANSYS)
Radar design, always a complex engineering challenge, becomes even more difficult when radar systems are placed on a vehicle moving through a complex environment, such as a typical city street. Designers must consider the performance of their radar technologies when installed in different fascias, placed in a variety of traffic conditions, and subjected to the extremes of weather and temperature. Fortunately, today’s advanced simulation solutions make all these engineering challenges easier to manage.
ANSYS has identified five specific areas where simulation can significantly impact the development of radar systems for autonomous vehicles:
• Antenna design
• Isolated radar simulation
• As-installed radar simulation
• In-environment radar simulation
• Closed-loop radar simulation
Accurate simulation of each of these aspects will be presented in this talk.
Christopher MacDonald (PTC)
Director of Analytics Business Development
In a world where physical and digital have converged engineers now have access to real physical data from inductive models to get insight for future designs. Being able to create a virtuous cycle that can be leveraged for the benefit of the product, both as designed and as operated, takes engineering to a whole new level.
In this keynote, Chris Macdonald will show how IoT Platforms enable companies to connect the physical and digital worlds to create new value and insights. New applications and services combine the data streamed from physical devices with intelligent digital models to solve problems throughout the entire product lifecycle. integration of product based computational models into IoT solutions will enable companies to quickly analyze operating conditions, rapidly triage and diagnose issues, predict future behavior, and optimize product performance.
Sudhir Sharma (ANSYS)
Smart connected products are enabling the digital economy, transforming organizations and entire industries. Autonomous vehicles, personalized medicine, and smart industrial assets, for example, hold the potential to deliver trillions of dollars of economic opportunity. Yet, creating and deploying cost-effective solutions to capture this opportunity requires changes to engineering product development processes and operations. In this session, we will share how leading companies are using engineering simulation platforms to develop smart connected products. You will learn how digital exploration, digital prototyping, and digital twins can help your enterprise maximize revenue and profitability through better product design, improved operations, and new services models.
Speed of Light Execution Enabled by Simulation
Sunil Sudhakaran (NVIDIA)
Director Hardware Engineering (Signal & Power Integrity)
Speed of light (SOL) execution is the cornerstone of NVIDIA’s business strategy and corporate culture. It is the minimum upper bound that is paradoxically within reach yet unattainable, and in the context of developing and productizing new technologies, it quite simply means as fast as possible. The majority of NVIDIA’s product lines are comprised of cutting edge processors which involve lengthy development cycles that include architecture, design, verification, fabrication, and post-silicon validation phases. Over-designing by simply adding cost to products is not feasible for competitive concerns, so design robustness needs to be cost aware. SOL execution requires excellence on the first try while being cost sensitive and anything else is considered to be a failure. One key practice is paramount for such an execution model: simulation.
Simulation is performed extensively throughout the design cycle at NVIDIA. This presentation primarily covers simulations related to system design including I/O interfaces and chip power delivery. NVIDIA is at the forefront of wireline link technology having productized a 11Gbps single-ended GDDR5x memory interface and the eponymous NVLink interface enabling GPU-GPU communication with data rates far exceeding existing PCI-Express technologies. Such high-speed interfaces require SI modeling with extreme attention to detail and unparalleled accuracy. SI feasibility simulations for these high-speed interfaces are carried out early in the development cycle to help define product architecture and specifications and continue until processor tapeout where the focus shifts towards verification. Furthermore, as systems are now power limited due to the end of Dennard scaling, a colossal burden is placed on the power distribution networks feeding into the chip due to the larger di/dt. Power integrity simulations are carried throughout the design cycle to guarantee power supply noise targets can be met.
In order to fit within development cycles and not become critical paths in and of their own, simulations need to be blazing fast. Using multiple CPU cores has provided appreciable yet incremental speedups for SI simulations. NVIDIA has shown across several applications that the time for GPU computing has come. Aforementioned mentioned SI & PI analyses need to leverage the power of GPU computing to provide non-incremental speedups to scale with accelerated product development cycles. Finally, the recently ignited AI big bang can most definitely revolutionize such simulation approaches.