Different organizations have set different autonomy goals for themselves ranging from
supervised L2/L2+ automated driving to unsupervised robotaxi applications (L4/L5).
Supervised automated driving itself has different tiers based on the driver’s engagements in
the driving task. Each use case puts forth a variety of constraints on autonomy sensing and
the compute architectures depending on the operational design domain, target market and
the level of autonomy. Examples of problems that an autonomy architect needs to find
answers to are – defining a sensor constellation ensuring adequate coverage in the region of
interest, determining field-of-view and resolution of the sensors, determining minimum frame
rate requirements for camera data ingestion, dependency on maps, homologation
requirements and consumer rating scores in the target markets. This talk takes us through
some of those use cases, posed constraints and methods followed to find answers to above
or other questions in the concept phase.