John Rushby, Program director, SRI Intern. Computer Science Laboratory, USA
http://www.csl.sri.com/users/rushby/
Bio : Dr. John Rushby is a Program Director and SRI Fellow with the Computer Science Laboratory of SRI International in Menlo Park California, where he leads its research program in formal methods and dependable systems. He joined SRI in 1983 and served as director of its Computer Science Laboratory from 1986 to 1990. Prior to that, he held academic positions at the Universities of Manchester and Newcastle upon Tyne in England. He received BSc and PhD degrees in computing science from the University of Newcastle upon Tyne in 1971 and 1977, respectively. His research interests center on the use of formal methods for problems in the design and assurance of safe, secure, and dependable systems. John Rushby received the IEEE Harlan D Mills Award in 2011 "for practical and fundamental contributions to Software & Hardware Reliability with seminal contributions to computer security, fault tolerance, and formal methods" and, together with Sam Owre and N. Shankar, the CAV Award in 2012 "for developing PVS (Prototype Verification System) which, due to its early emphasis on integrated decision procedures and user friendliness, significantly accelerated the application of proof assistants to real-world verification problems."
Title : Logic and Epistemology in Safety Cases
Abstract : A safety case must resolve concerns of two different kinds: how complete and accurate is our knowledge about aspects of the system (e.g., its requirements, environment, implementation, hazards) and how accurate is our reasoning about the design of the system, given our knowledge. The first of these is a form of epistemology and requires human experience and insight, but the second can, in principle, be reduced to logic and then checked and automated using the technology of formal methods. We propose that reducing epistemic doubt is the main challenge in safety cases, and discuss ways in which this might be achieved.
Pascal Traverse, Airbus Cockpit & flight ops. R&T Program leader, France
Bio : Pascal Traverse doctorate was on dependable computing and was directed by the late Jean-Claude Laprie and by Jean Arlat in LAAS. He graduated before as an engineer from ENSEEIHT and after visited UCLA to work with Al Avizienis.
He entered AIRBUS in 1985 and had the opportunity to participate to the design of the flight-by-wire system of the A320, the first time such a system was on a civil airplane. He then went on with the subsequent Airbus fly-by-wire developments, then leading the overall Airbus systems safety activities. Two of his patents are flying and some bits of Aviation Safety regulation are his authoring. He then focused on resilience: his own by working three years in the A380 Final Assembly Line and then Airbus Engineering’ as he moved this organisation closer to manufacturing concerns. He recently took the lead of Airbus research on cockpit and flight operations. With colleagues of “CISEC” (critical embedded system society of Aerospace Valley), he recently set-up a seminar on embedded critical systems.
Title : Embedded Systems Dependability (fly-by-wire)
Abstract : Safety is the priority in aviation. The reason is obvious: people's lives are at stake and this is ingrained in the industry (operators and manufacturers, regulation agencies).Instead of a wide description of the systems safety process, which is generally known and well documented, we will rather focus on a few important points, not always part of conventional wisdom. We intend to show that safety process is fully embedded in design, that the ubiquitous and magic number of 10-9 is only a tiny part of the solution, the multiple dimensions of the issue and the overall resilience of the process. This will be supported by examples taken from Airbus Fly-by-Wire system, typical of critical embedded systems and encompassing multiple dependability features. Architecture is highly fault-tolerant while fault prevention and removal is always in focus.
Sami Haddadin, Dr.-Ing. DLR, Germany
http://www.robotic.de/Sami.Haddadin
Bio : Sami Haddadin holds a Dipl.-Ing. degree in EE, a M.Sc. in CS from Technical University of Munich (TUM), and holds an Honours degree in Technology Management from TUM and the Ludwig Maximilian University Munich (LMU). He obtained his PhD with summa cum laude from RWTH Aachen. At the Robotics and Mechatronics Center of the German Aerospace Center (DLR) he acts as Scientific Coordinator ``Human-Centered Robotics’’. He is a lecturer of various robotics courses at TUM. In 2011 he was a visiting scholar at Willow Garage and Stanford University. His main research topics are physical Human-Robot Interaction, nonlinear robot control, real-time motion planning, real-time task and reflex planning, robot learning, optimal control, variable impedance actuation, brain controlled assistive robots, and safety in robotics. He was in the program/organization committee of several international robotics conferences and a guest editor of International Journal of Robotics Research. He published more than 75 papers in international journals, books, and conferences. Among other things, he received five best paper and video awards at ICRA/IROS, the 2008 Literati Best Paper Award, the euRobotics Technology Transfer Award 2011, the 2012 George Giralt Award, the 2012 IEEE Transactions on Robotics King-Sun Fu Memorial Best Paper Award. Furthermore, he was a finalist of the 2009 Robotdalen Scientific Award, IROS 2010 Best Application Paper Award, and 2012 SfN BCI Award.
Title : It is (almost) all about human safety: a novel paradigm for robot design, control, and planning
Abstract : Enabling robots for direct physical interaction and cooperation with humans and transferring the resulting technology to industrial and domestic real world applications is the primary goal of my research. For this, we developed new generations of impedance controlled lightweight robots (LWRs) at DLR, which are sought to act safely as human assistants in a variety of application domains such as industrial assembly and manufacturing, medical assistance, or house-hold helpers in everyone’s home. The last generation of the lightweight robot was recently commercialized as the KUKA LBR iiwa, which is considered as the first representative of a new class of robots: the so-called “soft-robots”. In the aforementioned applications, which aim for human environments, the primary objective is to ensure that a robot’s action does not cause human injury, even in case of malfunction or user errors.
Acting and reacting safely to unforeseen events in real-time on basically all levels of abstraction is crucial in both control and planning domain. For this, a robot needs to know in particular what injury it may cause due to its actions. In order to equip a robot with such knowledge, a systematic approach for human injury analysis and prevention had to be developed that bridges the gap between robotics and injury biomechanics. For quantifying what safe behavior really means the definition of injury, as well as understanding its general dynamics is essential. We approached the problem from a medical injury analysis and biomechanical point of view such that the relation between robot mass, velocity, impact geometry and resulting injury in medical terms can be found. Due to the achieved generality of our results, this methodology has also found its way into current international standardization effort.
Algorithmically, so called “biomechanically safe velocity” and high-speed collision detection, tightly entangled with appropriate reflex reaction build the robot’s real-time “safety core”. The “biomechanically safe velocity” algorithm ensures truly safe robot velocities by evaluating in real-time the potential injury risk emanating from the robot’s inertial and surface properties and shaping its velocity such that accidental collisions with the human body remain subcritical. Collision detection and cascaded reflex reaction schemes let the robot then react safely after a collision was sensed. Thus, subsequent potentially hazardous situations are avoided if possible and overall, entire injury prevention even in case of unforeseen collisions becomes possible. This control core builds a reliable foundation for novel real-time motion planning and collision avoidance schemes that bridge the gap between planning and control, while also exploiting the aforementioned injury knowledge in order to generate “human-safe” trajectories.
The consecutive encapsulation of all real-time algorithms into an (action-behavior) based “control core” builds then a powerful basis for human-friendly task planning, as the large variety of new features become part of a safety-oriented programming model. In this context, I will also report on our recent results in dynamic programming paradigms and provably optimal, safety-oriented planning algorithms that finally make it possible to bring novel robot control algorithms, human safety, and human-friendly task planning into alignment.
Overall, I will argue in my talk that the described human-safety oriented design, control, and planning paradigm contributes significantly to solving some fundamental problems of nowadays robotics and let “soft-robots” become a commodity in our near-future society. I will also illustrate the relevance of the field and the according technology with selected real-world implementations of our concepts such as advanced manufacturing solutions at automobile companies. Their aim is to achieve flexible production lines that involve human-robot co-working and are free of safety cages.
