Smart systems are a hot topic these days. Enabled by continuously shrinking electronics, we can now add software to almost anything, from thermostats to autonomously driving cars and self-regulating energy networks. But is a smart system also trustworthy? Will it ever be smart enough to be trustworthy? Self driving cars for example use artificial intelligence to mimic the behaviour of an experienced driver.
Clearly, this is a hypothetical driver as human drivers make mistakes all the time. And how would you feel when your car updates its driving software while driving? How smart is a self-driving software that keeps on speeding after it detected six times in a row that the driver was not paying attention to its warnings? How smart is it to have a dishwasher talk to the internet so that it doesn't work when you have no connection? Of course, this type of smartness was introduced by smart people. The issue is complexity, made worse by legacy developed at a time when trustworthiness was a second thought. In the real world the complexity is immense and the law of Murphy is king. The question is how can we make things simultanously smart and trustworthy? Another question is whether such a smart system should mimic a human brain? Trustworthy means predictable, safe and secure which are all aspects for which humans have a very bad reputation. The road to salvation is a systematic one whereby unavoidable errors and faults are taking into account at every step of the development. Fault tolerance and resilience are key.
To enable intelligent cars, it is critical to upgrade their perception capabilities. Depth is one of the key senses required. In this presentation we will give an overview of the different depth sensing technologies and present the progress on SoftKinetic’s own DepthSense® technology and its usage inside the car for gesture control and in cabin monitoring.
Last year, a reported 1.000 European cyclists and pedestrians were killed in collisions with trucks, leaving many more seriously injured. Progress on reducing casualties among vulnerable road users has been particularly disappointing. The problem lies mainly in the blind spots alongside heavy vehicle, which render other road users often invisible to the driver. Especially in urban areas, collisions between trucks and cyclists remain a grave concern, responsible for more than half of all fatal cyclist accidents. With expanding urbanization and the rapidly increasing popularity of cycle commuting, concerns regarding cyclist safety are growing worldwide.
Too long, national and EU policy makers have focussed on improving the indirect vision of trucks by means of mirrors and cameras in an attempt to reduce the number of blind spot casualties. However, during complex manoeuvres it is virtually impossible for a driver to keep at all times a correct and complete overview on the position of all road users in the vicinity of his vehicle. This is aggravated by poor vision conditions and the distractions of hectic urban traffic.
Inspired by upcoming legislative changes, technology companies are racing to deliver new smart safety technology, capable of constantly monitoring the driving environment and warning the driver when a collision is imminent with pedestrians or cyclists within the vehicle's danger zones. Such systems not only leverage human awareness, but also integrate into a much broader Smart City mobile sensor platform that aggregates vehicle data and collision trends, which is of paramount importance to help cities identify effective road safety measures.
In this presentation, we will address some of the technology challenges faced in the development of smart driver assistance systems to avoid blind spot collisions involving vulnerable road users.
Autonomous driving has important implications and challenges for the engineering of the car systems and its software and electronics.
Cars will be instrumented with a growing number and variety of sensors giving a tremendous amount of data and information. Software and hardware gets increasingly complex and much more integrated. For safety, a massive amount of scenarios and cycles will need to be validated and verified. Combining traceability and agility is key to manage a liability shift from driver to OEM, requiring well documented development processes, while there is new technology coming to market every day. All these new challenges need to be tackled while not compromising on fuel efficiency, comfort and drivability.
This requires a paradigm shift and a rethink of the complete vehicle development process.
This presentation will give an overview of these challenges, will explain Siemens’ view and strategy, and will go more into depth on the possible implications on engineering tools and design processes. It will also indicate potential topics of cooperation with Flanders industry and research institutions.