Autonomous vehicles, whether to transport people or for freight delivery – can bring huge disruption to our lives both in business and society. There are many benefits to be had, including a decrease in accidents resulting from human error, a reduction in costs and environmental impact of transportation, a reduction of the time that is currently spent driving, and access to more users.
In this regard, it is clear that the most pressing obstacles must be overcome in order in order to realize this vision:
- The assurance of software and systems: How can we identify and demonstrate the proper degree of acceptance?
- Connection and Sensing: What can we do to ensure that there is a good relationship between a car and its surroundings?
- The Judgement: Can an automated system use judgment?
- Architectures to manage complex systems: What can we do to deal with the system complexity that results?
- Verification and validation: How much testing are we required to do, and how do we get it?
In the context of this market and this study, a variety of possibilities and strategies to meet these obstacles can be analyzed:
The final acceptance of autonomous cars will be a collective and political choice; therefore, the people involved are required to make clear their decision-making process and the reasoning for their decisions. The assurance that is employed by other industries is likely to be difficult to procure.
The complexity of the driving environment will require new sensors as well as new communication channels, along with more sophisticated methods to understand and analyze the data.
- The process of implementing decision-making processes should take into account the following:
- A proper separation of the responsibility among manufacturers, operators, and other partners that is based on precise technical specifications instead of abstract objectives.
- The capability to modify and correct decisions made as time passes.
- The importance of human-machine interactions will require user-centric design methods to be implemented.
Autonomous systems can be extremely complex. Architectural strategies will be required to reduce costs and also to make the assurance of safety feasible.
Whatever targets for assurance are established in the future, the complexity of the vehicle and their surroundings makes testing difficult and therefore:
- Test strategies that are able to support huge and well-defined testing programs are needed.
- Evidence from a diverse array of assurance methods (not just testing that is dynamic) will have to be utilized.
