Safety can be improved by the three factors: infrastructure, vehicle and driver. Recent accident analysis has shown that the driver is responsible for 97% of all accidents.
Two measures for driver improvements can be considered:
- Improving reliability of the driver e.g. by feedback monitoring.
- Increasing substitution of the driver both on “middle & long-term” - corresponding to higher degree of automation with first demonstrations like in Chauffeur and on “short-term” – meaning intervention in critical scenarios.
HAVEit mainly aims to long-term substitution (typically a few seconds), thus leading to higher a higher level of automation and furthermore to the transition to short-term substitution.
With higher level of automation, additional challenges are coming up, i.e. the question "how to optimally keep the driver in the loop" and "how to ensure optimal and safe termination of the automation and stepwise transfer of the driving task from the co-pilot system back to the driver". To cope with these challenges, HAVEit will care for the link between automation function, vehicle and driver by means of a decision control to guarantee safe and adaptive task repartition in the joint system.
Highly automated driving can strongly support traffic safety, as it can address the main causes for accidents as shown in the following sections.
- Highly automated driving for the improvement of overall safety
Higher integration and more reliable and driver sensitive support and intervening safety functions can improve road traffic safety. A significant step towards higher road safety can be achieved by combining intervention and substitution models within a cost-effective platform. HAVEit will upgrade and integrate the safety proven and failure tolerant SPARC architecture to meet the requirements of highly automated driving.
Long-term "Highly Automated Driving" has the following meaning in HAVEit:
The key actor for safe driving must be the driver. Everything has to be done to optimise her or his performance. An automation centred on the driver is an important tool to achieve this goal. Taking into account that the need for assistance strongly depends on the varying performance level of the driver, the need for a variable task partition between driver and automation becomes obvious. Higher degree of automation in this context means to support the driver in monotonous driving tasks (e.g. queuing on crowded motorways) as well as in highly demanding tasks (e.g. automated merging into the traffic flow). Automation must be designed in a way that different degrees or stages of support can be flexibly produced (ranging from mere warning up to a temporary auto-pilot etc.).
One of the key issues is the definition of the optimum way for sharing the driving task between the driver and the highly automated system. If the driver is being allowed to be out of the loop for a few seconds, he or she needs to get a few seconds time to get back into the loop and to react properly (i.e. to continue driving) in case the highly automated system faces a scenario that cannot be handled automatically. Current state of the art systems and other research systems warn the driver in such cases and just switch off, thus leaving the driver alone in the critical situation. HAVEit will follow a novel approach to overcome this problem. A step-wise strategy (starting sufficiently in advance of an up-coming situation that might not be handled automatically) will be developed to give the driving task step by step back to the driver, to ensure that the driver is really capable to suitably react to the driving situation. Due to its safety relevance, the optimum task repartition in the joint system driver / co-pilot has been identified as the first key objective of HAVEit and is considered as a horizontal activity.
The multi-stage concept for optimum task repartition in the joint system will be made available to the vertical application activities aiming at different innovative cutting edge safety and comfort ADAS. The next generation ADAS systems have been identified as the second key objective.
Systems aiming at higher degree of automation and on safety definitively need to include the possibility of intervention to avoid a collision or mitigate a crash in cases the driver is for a dedicated moment out of the loop. To take this safety requirement into account, at least one of the vertical applications to be developed will integrate an autonomous emergency braking function into the highly automated driving features for public traffic.
- Highly automated driving to improve comfort and to pave the way for safety
A first step for the creation of a co-pilot system to support the driver has been done in the PEIT project: A vehicle platform has been generated being able to place an extended stability control (ESP) as execution level manager receiving a vehicle command from the driver (as intelligent command level) and to improve reliability and reactive active safety quality.
A second step has been achieved in the SPARC project with the integration of a safety-relevant human oriented architecture on top of that. The ADAS are now able to intervene for crisis management (anticipation, interaction and recovery) by improving the driver's command before sending it to the executive level as vehicle command. The proof of concept has been successfully validated in SPARC.
When trying to industrialize this safety approach, it has been recognized, that on the one hand most end-users do want to have safety, but are willing to pay for it only minimally. On the other hand, with the full pressure of the market, OEMs cannot integrate safety for free. Thus, safety has to be brought indirectly and for free by means of integrating also other functionalities relying on the same platform. Right now, end-users accept to pay for comfort features and fuel saving.
On that basis, a safety oriented technology will get wide acceptance if it results in cost or packaging advantages compared to the current state-of-the-art or if it leads to new valuable functionalities for the end-user at acceptable additional cost.
Both directions are in the scope of HAVEit: By inverting the problem, i.e. by integrating automation capacity and fuel saving possibilities with a fully redundant, failure tolerant safety vehicle architecture the massive deployment of the cost-efficient safety architecture on the road will be prepared.