Intros
Jean: I wanted to present what
Geotab is doing around SmartCities
... we are a traditional telematics company, although we do
things somewhat differently
... core business is fleet management but realizing there are
many derivative value
... we try to convince our customers that their vehicles are
essentially smart sensors on wheels
... additional sensors are adding to the many potential use
cases
... everything I will present are about how we are leveraging
this data for DOT focuses
... we have a number of core strengths including security
... we are addressing high value use cases, trying to get at
ADAS data but also many common existing data points
... we have approximately 1.4M vehicles producing 3B data
points per day
... this map represents one day of data collection
... I mentioned we developed some capabilities around smart
city use cases
... we have broken it down into five segments, a few of which I
will cover
... we are doing some really interesting things in California,
working with various state agencies and municipalities
... we are starting to collaborate with their EPA agency,
looking at heavy duty truck activity within their
jurisdictions
... if you can approximate volume of gas or diesel at a
specific temperature, approximate the impact on regional air
quality down to particulate amounts
... what you are seeing is a model developed with CARB and
environmental defense fund
... we are approximating air quality impact from heavy
vehicles
... we are starting to equip public vehicles with air quality
sensors
... creating essentially roaming air quality sensors
... you need mobile, roaving collection as fixed points do not
provide adequate coverage
... we also work with various state/regional DOT and public
works departments to evaluate condition of their road
surfaces
... to help develop road maintenance plans again leveraging
vehicle data, adding sensors like accelerometers
... from a safety perspective from the vehicle itself and our
own sensors, we get aggregate information from 100s of
thousands of vehicles to identify hazardous driving areas
... which intersections or areas have dangerous areas (based on
data sampling)
... we are starting to predict where future collisions will
occur. first responders are proactively dispatching based on
these findings
... this is based on machine learning
... many of the vehicles we connect today have built in
temperature sensors and that is useful for identifying
potential freezing conditions
... we can identify if anti-lock brakes or TCS engage to spot
black ice
... traffic flow is another area we get into although other
companies focus more extensively on it
... we can identify commercial speeds to help coordinate
traffic lights on a certain corridor
... we feel the traffic slowdown is exaggerated by heavy
vehicles and they should be prioritized as a result
... we can identify areas with parking challenges. we can
identify curb space and double parking frequency in NYC for
example
... last is EV charging. we are working with state of
California, and others, for private and commercial fleet
electrification
... where additional charging stations should be deployed. you
need to first understand traffic flows and where they tend to
charge
... we have some locations this is not a concern, for instance
high volume of EV in the morning but typically vehicles have a
full charge
... retroactively we were able to identify the epicenter of the
Mexico City earthquake based on the accelerometer of our
vehicle in the area alone
PatrickL: questions from the
group?
... please do share your slides so we can redistribute
internally
Jean: I will send the slide deck directly
Gunnar: I would be interested when these needs can help influence the standards work
Jean: good point, we needed to change the way we are collecting data
Gunnar: I really liked to hear about predictive accidents
Don: I work for Eonti which is
providing PKI for OCF
... I am chairing OCF automotive activity
... including mapping VSS data model, the previous incumbent
had done considerable work but stopped six months ago
... I want to continue that work and have reached out to Gunnar
and Ted to see where things stand as I want to expand the scope
beyond the use cases we were working with
... short answer is both groups (GENIVI and W3C) are expanding
scope
... one of the first things we want to address is
security
... some devices work in more than one ecosystem at a time
which is a security challenge
... the initial work within OCF was coming from a smart home
view, bridging entertainment space to vehicle
... focusing on IVI, mapping the various entertainment
aspects
... here it was constrained to that set of functionalities. at
CES a couple years ago they demoed a car as a connected
thing
... there are other trends and we have additional motivations,
ridesharing, electification and AV
... list on this slide is more of a wish list going well beyond
IVI, V2i, V2V, Smart Cities...
... a vehicle for an individual is connected to the home, see
what should be added to data models and specs
... a number of these are same as Jean mentioned
... vehicles talk to other vehicles, infrastructure etc. a key
area from US DOT is PKI management
... car could be at home, connected to a charging station.
there is an OCF security model as part of the vehicle/charger
connection
... there are already a number of CA from utilities with its
own security model for communicating between the charger and
utility
Ted: at W3C TPAC this automotive group met with a number of other W3C groups including the Linked Building Data Community Group. if it makes sense I can do an intro
Don: they more commercial or residential building focused? Either way, it would make sense
Ted: commercial but unsure, they may also be working on homes
Don: are you aware of the Fair Hair alliance? started from a Scandinavian company and we are liaising with them
Gunnar: I think you might be talking about Norwegian Herald
https://www.fairhair-alliance.org/
Jean: we are following US DOT/NHTSA PKI proposal and wondering if it will go forward or not
Don: I don't believe so. I think the issue they are focusing more is on spectrum allocation for DSRC ten years ago and now 5G is starting to compete
Jean: so the belief is this will be the security layer across both?
Don: that is one thing under consideration, regardless of the transport. the PKI/Security Credential Management System work will go forward, it is independent of/agnostic to the transport layer discussion about spectrum allocation.
PatrickL: reaction so far to
Ulf's proposal was positive and agreeing with the logical
separation
... comments were going toward the direction of not moving
together the language and words from HTTP but keep a more pure
interface
... any other thoughts on summarization?
Gunnar: my interpretation was
slightly different but want to hear from others
... my concern is what do we mean by transport and
interface
... if you try to bring to a common interface it might be
suboptimal for certain uses
... the mapping to HTTP REST might not be perfect. we should
leave it open
PatrickL: this is in alignment with the discussion at our face to face meeting
Ulf: short term question is whether this is good enough to put into github to be able discuss further
PatrickL: from my perspective, yes
Ulf: alright, I will put it out
there. we should also keep Patrick's document there as it is
great about pulling our various past discussions and
conclusions
... I will put these two files in where we will write the
actual specification
Gunnar: if you can read and
respond to my email beforehand that would be appreciated. I
will accept if the agreed direction is different
... perhaps I was too late to the discussion
Ulf: all input is good
... entering these two documents is not etching in stone but a
starting off point
... any part is open to discussion and change
... I read, although quickly, what you wrote and understand it
but not what the impact/result to the document itself
... once it is there perhaps you can provide your perspective
further
Gunnar: one document is concerning data model and I understand we potentially want to allow for more than one transport
Ulf: data model is coming from VSS
Gunnar: ok let's try this. I was concerned the interface might be difficult for the various things we want to do but let's try it
PatrickL: any other comments or objections?
Paul: I agree
Ulf: I will do it within a few days
PatrickL: Magnus, I think we will have to defer on your security topic until next time
Magnus: mostly want to see if there is interest, this company is willing to provide time
Ted: short version is yes, we're interested and trying to align some other people on security/privacy as well. this is Ryan's area of expertise for instance
Gunnar: intro or we have something for them to review? seems premature for latter
Magnus: I will continue discussion with them
Patrick: anything else?
[adjourned]