W3C

Automotive data task force Teleconference

18 Apr 2019

Attendees

Present
Ted, Benjamin, Harjot, Armin, Glenn
Regrets
Chair
SV_MEETING_CHAIR
Scribe
ted

Contents


Summary of data task force

Ted: most of our focus has been on learning more about the VSSo ontology work, what we have dubbed 'data contracts' and consent/policy

Harjot's Data Contracts document

scribe: transportation data workshop date for week of 9 September in Palo Alto with W3C Auto WG F2F to immediately precede or follow the remainder of the week
... as mentioned, I see some of the topics we have been exploring to potentially be within scope of potential new activity resulting from the workshop
... specifically additional ontologies including metadata being discussed in data contracts context and privacy/policy work
... expecting more semweb ontology/vocabulary expertise to be part of that group
... sampling methodology we saw as potentially addressable with W3C PROV ontology specs. we can investigate further to make recommendations to this unformed group

Consent and Policy

Ted: regarding consent and policy we have been waiting for Armin and his colleagues to further LPL before trying to figure out how to apply it
... he indicated they are at a point where we can start looking at trying to use it against a real data privacy policy
... I reached out to a three OEM for their data policies, have not received any yet and will be discussing in person with at least two if not the three between now and f2f in Munich

Armin: started to looking into the requirements and putting the language together
... I should be able to do a presentation on the next call on what we would require
... would like to share what I am thinking of

Benjamin: are we planning something for the next f2f meeting or would it be auto WG focus or with broader Genivi community

Ted: yes to the WG@@yes, we should have an agenda item where this task force reports to the WG and also for inclusion in your Genivi presentation

Benjamin: I will send you what I have and would appreciate the comments

Glenn: for Gen2 a developer can create an app for a mixed vehicle fleet, provided those OEM make them available in their marketplaces

Benjamin: what do you mean by mixed fleet?

Glenn: a fleet comprised of vehicles from multiple manufacturers, example being a typical rental car company
... challenge is the fleet manager would want to be able to access data and interact with vehicle consistently

Ted: I am considering the policy work as a prerequisite for delving further into security concerns, need to know what sorts of interactions we are trying to protect first and the fundamental security considerations in VISS & Gen2
... a couple things we have previously decided to raise as issues in either W3C Gen2 or Genivi VSS repos

Wake events

Ted: Wake events and heartbeats was raised to the W3C Auto WG and there was skepticism about whether OEM will wake the service and head unit or vm where apps will reside

Wake events and heartbeat #300

scribe: while there are sound business arguments to make still for OEM to consider this, there are no provisions being discussed in Gen2 spec
... we can revisit as Gen2 advances or look outside the spec as it may not need any special provisions within it to make this possible. service does not need to know rationale for why it is running, whether it was awakened by some external control for sanctioned apps
... I could see a policy directive can indicate to a capable vehicle an application will want routine heartbeat or event based trigger to start service and vm app resides on to poll data

Benjamin: what does it mean for us to make a business case? do we just provide use cases and markets that would be impacted?
... how much should we do

Ted: W3C groups produce specs as we are aware but also can produce guidelines and best practices. we can collect use cases and business rationale for wanting wake events for OEMs to consider

Harjot: are you asking for specific use cases for heartbeat functionality

Ted: data your care about and problems trying to solve

Harjot: GPS for location, accelerometer are used to tackle theft or towing
... if we sense motion, wake up and continue login. same with GPS. we verify location every hour for first 24h
... other data points from vehicle are fuel level, ensure it hasn't changed (gas being siphoned or leaking)
... doors and trunks opening at next hearbeat would be helpful for unauthorized or scheduled access
... there are plans for instance to be able to remotely open trunks for receiving deliveries
... we are seeing this with smarthome locks in North America for now

Glenn: another example use case related to fuel is for eg California wildfires to be able to let the firefighters know volume of volatile fuel is in a given area

Ted to start wiki of business & use cases

Precision

Ted: for given signal it would be nice to have the degree of precision (*if known) coming from the underlying sensor, is it accurate to within eg 5C?
... we thought this belongs directly in the data model itself and should raise an issue in Genivi VSS repo
... we haven't but there are a couple nearby issues, 44 marked as a duplicate so we can either create another or expand on #41

Need to support fraction of inches? #41

Altitude with double precision? #44

Harjot: I think a new one would be more appropriate

Ted to start it

Data Sampling

Ted: a couple meetings ago we were discussing data sampling and how it may too belong in the data model itself
... in addition to cloud centric use cases for knowing how data was collected, it would be useful for apps in the vehicle to know how the signal the service is exposing is produced: regular time interval, event trigger, new peak value
... apps running on the vehicle and sending data to the cloud can use the same metadata attributes to convey how it did collection
... we want to figure out what to include in the data model but need to try to avoid scope creep

Benjamin: this is worth bringing to the Auto WG, know what to include and not, good practices to follow for modeling vehicle data and not just signals
... there is always this hypothesis about how we use those models, needs vary and have different requirements and different desired metadata
... precision, access, extend entries in VSS leaves is one solution
... Patrick wanted a more open data model where those attributes can be included in another way, this would be worth an agenda at F2F
... we do not know everything at this point but can bring our observations
... we can maybe reduce the scope as well for both us and the WG
... YAML vspec from VSS with mapping to whatever data source you have and create a spec with new entries

https://www.w3.org/auto/wg/wiki/Main_Page#Upcoming_F2F_Meetings

also suggest registering for Genivi's open day on Wednesday

[adjourned]

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version 1.154 (CVS log)
$Date: 2019/04/22 14:41:35 $