Sunday, April 28, 2013

Autonomous, not driverless

Paul Leroux
I don't know about you, but I'm looking forward to the era of self-driving cars. After all, why spend countless hours negotiating rush-hour traffic when the car could do all the work? Just think of all the things you could do instead: read a novel, Facebook with friends, or even watch Babylon 5 re-runs.

Unlike Babylon 5, this scenario is no longer a page out of science fiction. It’s coming soon, faster than many imagine. That said, the story of the self-driving car still has a few unfinished chapters — chapters in which the human driver still has an important role to play. Yes, that means you.

As I’ve discussed in previous posts, the fully autonomous car is a work in progress. In fact, some of the technologies that will enable cars to drive themselves (adaptive cruise control, forward collision avoidance, etc.) are already in place. Moreover, research suggests that these technologies can, among other things, improve traffic flow and reduce accidents. But does that mean you will soon be able to sit back, close your eyes, and let the car do everything? Not quite.

Evolution, not revolution
If you ask me, Thilo Koslowski of Gartner hit the bull's eye when he said that self-driving cars will go through three evolutionary phases: from automated to autonomous to unmanned. Until we reach the endpoint, we should pay heed to the words of Toyota's Jim Pisz: autonomous does not mean driverless.

If planes can do it…
Some folks hear this and are disappointed. They point to auto-pilot technology in planes and ask why we can’t have driverless cars sooner than later. The argument goes something like this: "It's much harder to fly a plane, yet we have no problem with a computer handling such a complex task. So why not let a computer drive your car?”

If only life were so simple. For one thing, automakers will have to make autonomous cars affordable — doable but not easy. They’ll also have to negotiate a variety of legal hurdles. And in any case, driving and flying have less in common than you might think.

When you drive, you must remain alert on a continuous basis. Lose your attention for a second, and you stand a good chance of hitting something or somebody. The same doesn't always hold true in flight. When a plane is cruising at 30,000 feet along a proscribed flight path, the pilot can avert his or her attention for 5 seconds and incur little chance of hitting anything. In comparison, a driver who becomes distracted for 5 seconds is hell on wheels.

And, of course, auto-pilot doesn’t mean pilot-less. As Ricky Hudi of Audi points out, pilots may rely on autopilot, but they still retain full responsibility for flying the plane. So just because your car is on auto-pilot doesn’t mean you can watch YouTube on your tablet. Bummer, I know.

An alarming solution
Source: Modern Mechanix blog (and yes, that should 
read Frankfurt)

All of which to say, the driver of an autonomous car will have to remain alert most or all of the time — until, of course, autonomous vehicles become better than humans at handling every potential scenario. Now that could happen, but it will take a while.

It seems that someone anticipated this problem in the early 50s when they invented “alarming glasses” — take a gander at the accompanying photo from the August 1951 issue of Modern Mechanix.

Scoff if you will, but a kinder and gentler form of this technology is exactly what autonomous cars need. No, I'm not suggesting that scientists find a better way to glue wires to eyelids. But I am saying that, until cars become fully and safely autonomous, drivers will need to pay attention — after all, it’s tempting to drift off when the car is doing all the work. And, indeed, technologies to keep drivers alert are already being developed.

Pre-warned means prepared
Mind you, it isn’t enough to keep the driver alert; the car may also need to issue “pre-warnings” for when the driver needs to take over. For instance, let’s say driving conditions become too challenging for the car’s autonomous mode to handle — these could heavy rain, a street filled with pedestrians, or an area where lane markers are obscured by snow. In that case, the car can’t wait until it can no longer drive itself before alerting the driver, for the simple reason that the driver may simply take too long to assess the situation. The car will need to provide ample warning ahead of time.

The more, the better
That cars will become autonomous is inevitable. In fact, the more autonomous, the better, as far I'm concerned. Research already suggests that technologies for enabling autonomous driving can, in many cases, do a better job of avoiding accidents and improving traffic flow than human drivers. They also seem to do better at things like parallel parking — a task that has caused more than one student driver to fail a driving test.

But does this all mean that, as a driver, I can stop paying attention? Not in the near future. But someday.

Thursday, April 25, 2013

Goodbye passwords, hello biometrics

Let's face it — passwords suck.

Every day we have to recall all manner of alphanumeric combinations for bank PINs, network log-ons, corporate email, social networking, and e-commerce. According to Microsoft Research, the average user types eight passwords per day.

During a talk at last year's SAE Convergence, Joseph Carra from the US Department of Transportation said, “Passwords have to go” ... a breath of fresh air for those of us who rely heavily on the "forgot password" option. The stage is set, according to Carra, for biometrics to replace passwords in the vehicle.

Using biometrics for driver preferences is nothing new — my favorite example is a car seat that can identify you by the shape of your butt — but using them to replace passwords makes perfect sense.

Ultrasound fingerprinting, iris scans, facial recognition, signature dynamics, voice recognition, keystroke dynamics, hand geometry, skin patterns, and foot dynamics are already being used in enterprise security, law enforcement, border control, ATM transactions, and so on. And second-gen biometrics promise to pump up the Sci-Fi factor with neural wave analysis, electro-physiological biometrics, skin luminescence, body odor, and so on.

Many technologies eventually find their way into the car after becoming popular elsewhere — mobile telephony, media players, GPS navigation, etc. I can’t think of too many world-changing technologies that got their start inside the car. But given the innovative trajectory of today’s auto industry, that may be about to change.

Tuesday, April 23, 2013

A matter of context: How digital instrument clusters can enhance the driving experience

I always drive a manual, so checking the tachometer in my car’s instrument cluster has become second nature to me. But while I have a personal interest in what my cluster displays, why would a software company like QNX be interested in instrument clusters? After all, most clusters use physical gauges and relatively little software.

The answer, of course, is that automakers are starting to migrate to digital instrument clusters, which replace mechanical gauges with virtual instruments rendered on an LCD display. In fact, Jaguar and Land Rover, who are pioneers in this market, have been shipping QNX-based digital clusters since about 2010. Here, for instance, is a photo of the digital cluster and dash in the latest Range Rover:

So why use a large LCD display instead of mechanical gauges? For one thing, you can attract early adopters who always want the latest tech and who see large 3D displays as cool. But more importantly, a digital cluster can provide an experience that is both personal and adaptive — personal because consumers today want to control the UX (just as they customize their smartphones) and adaptive to help the driver in a variety of traffic situations.

Context matters
In the latest QNX technology concept car, for instance, the digital cluster can re-configure itself to display a 3D rear view camera to help with parking. Saab pursued similar ideas a few years ago with a context-based cluster that avoids loading the driver with too much information during night-time driving.

It will be interesting to see who takes this to the next level with an adaptive HMI that takes speed, location, and driving conditions into account. For instance, driving at high speed on a German Autobahn differs immensely from driving at low speed on a busy downtown street with lots of pedestrians and intersections. These two scenarios place different demands on the driver, and a digital cluster could adapt accordingly.

On the autobahn, the cluster could increase the size of the speedometer and tachometer to make them easier to see, while hiding other information that isn’t currently needed. (The cluster would, of course, still display any necessary warnings, such as high oil temperature.) In the city, meanwhile, the cluster could replace the tachometer with pedestrian warnings to improve the driver's situational awareness.

Also, think of a car that supports both automatic and manual gear-shifting. A driver who prefers automatic might not be interested in a tachometer, whereas a driver who shifts manually will want to see a RPM readout to optimize gear shifting. A digital cluster could accommodate both preferences.

For safety’s sake
What does it mean from a safety perspective to include a large display and its attendant electronics in the car? A malfunctioning digital cluster can’t directly kill or injure, but it could give false indications that may lead to an accident. That is why automakers will likely have to address ISO 26262 requirements for their digital clusters.

So what is ISO 26262? It’s a standard that focuses on functional safety in cars and other types of passenger vehicles, with the goal of avoiding or controlling system failures. It is similar in content and purpose to the IEC 61508 functional safety standard, to which two QNX OS products have already been certified. Read our previous posts (here and here) for more information on ISO 26262.

Massive arrays
When it comes to digital clusters, I’ve only scratched the surface. For instance, cars are becoming massive sensor arrays that generate tons of data. By leveraging this data, reconfigurable clusters could display contextually relevant information, such as highlighting a person in your path, an accident up ahead, or the current speed limit.

And from the automaker’s perspective, a digital cluster could help reduce costs by allowing the same hardware to be used across multiple vehicle lines; in many cases, only the graphics would need to be “reskinned.”

Emil Dautovic is an automotive business development manager at QNX Software Systems, where he is responsible for the European automotive market.

Wednesday, April 17, 2013

Finally, I can throw away my 8 tracks

Okay, maybe I'm not old enough to have owned 8-track tapes. But I do remember that my uncle had an 8-track player in the dash of his station wagon when I was a kid, and I am old enough to have owned a car with a cassette player.

Music has been fundamental to the driving experience for about as long as cars have been on the road. Terrestrial radio dominated forever, supplemented by tape and then CD. XM radio came along in 2001 and connecting your iPod started to show up in the late 2000s. That's 5 formats since the Model T was introduced in 1908 (okay, so it didn't  have a radio) and 3 formats in the first 90 years.

Now, with connected cars becoming a reality, the rate of change is shifting into overdrive. Want Pandora – check. Want to listen to the top alternative radio station in Dublin while driving in California – check. Want to keep listening to your Songza programming as you move from the house to the car – check.

Today's announcement from QNX and 7Digital adds a whole new dimension. Having 7Digital in the car will unify the music ownership experience across the big three: car, pocket, and home. Want to listen to your own music programming in the car – check. Want to buy a song you just heard on that Dublin radio station – check.

Read the press release for details. And when you're done, check out the 7digital blog.

Monday, April 15, 2013

Enabling drivers to interact safely with applications and services

Since February 2011, QNX Software Systems has been leading an international standards effort to help drivers interact safely with applications and services. And not just apps on phones, but apps running in the cloud, in roadside infrastructure systems, in the car itself, and other locations.

If you jump to the end of this post, you’ll find a list of use cases being targeted by this effort. For now, let’s look at Use Case 2, Scenario A (arbitration of external message), which illustrates how we are working towards a comprehensive framework for managing distraction and workload.

Keeping priorities straight
In this user scenario, a navigation maneuver is given priority over a social media status update message. The blue call-out boxes indicate where the ITU-T recommendations under development can enable safe interaction between the driver and applications. For instance, ITU-T recommendation G.SAM will define mechanisms for prioritizing navigation, while G.V2A will define the communications interface between the app and the driver-vehicle interface (DVI), and P.UIA will recommend characteristics of the auditory social media message.

Remember that the focus here isn't on how to implement social media in the car, but rather, on how best to manage workload and distraction.

Giving a navigation maneuver priority over a social media status update message

Often, I am asked how this effort differs from the MirrorLink standard being developed by the Car Connectivity Consortium. The simple answer is that MirrorLink addresses only some of the use cases listed below. For instance, the scope of MirrorLink is limited to applications and services running on nomadic devices. Furthermore, adaptation of the driver-vehicle interface and external applications and services in the current MirrorLink solution uses a simple two-state approach, driving or not driving, which limits the ability of the vehicle to control the timing and modality of communications with the driver. Also, MirrorLink doesn’t adequately address arbitration or integration of communications with all external applications and services.

In for the long haul
At QNX Software Systems, our aim is to:
  1. Work with the relevant parties to identify solutions to the problem of technology-related driver distraction and workload. These parties include automotive, telecommunications, and consumer electronics organizations; standards development groups; academia; and government agencies.
  2. Determine which aspects of the solution should be standardized, then help drive this standardization.
  3. Align QNX product roadmaps as solutions develop.
To be clear, this is a longer term strategy that will take years to realize. Both the standardization process and the time it takes to deploy technology in vehicles must be factored in. Therefore, we are also pursuing shorter term solutions, some of which I hope to cover in future posts.

The end of the beginning
The first major milestone in this effort was achieved at the closing plenary of the ITU-T Study Group 12 meeting, held on March 28 in Geneva. Here, the final report and 4 deliverables of the ITU-T Focus Group on Driver Distraction were approved. There was also approval of a liaison statement communicating these results to a large list of organizations working on this topic.

This marks the end of the focus group, but is really just the beginning for QNX and ITU-T efforts in this area. In future posts, I will explore various aspects of this comprehensive strategy.

Use cases and user scenarios targeted by ITU-T recommendations

Use Case 1: Interaction with external application/service
   a) Application on nomadic device
   b) Application on cloud-based server
   c) Downloaded Application
   d) Broadcast of roadway information
   e) Tethering
Use Case 2: Arbitration and integration of external message
   a) Arbitration of messages
   b) Integration of messages
   c) Both arbitration and integration of messages
   d) E-call
Use Case 3: Negotiation of network Quality of Service (QoS)
   a) Application selects network
   b) Application suspends interaction
   c) Application availability due to roaming
Use Case 4: Management of multiple dialogues
   a) Opening/closing an application
   b) Switching between applications
   c) Interaction with background application
Use Case 5: Adaptation of DVI (driver-vehicle interface) and external applications/services to driver abilities
   a) Driver with disability
   b) Dynamically changing driver capabilities
   c) Detection of impaired driver state
Use Case 6: Adaptation of DVI and external applications/services to roadway situation
   a) Driver busy notification
   b) Delay of message delivery in demanding driving situation
   c) Change message format based on road conditions
   d) Interruption of driver interaction
Use Case 7: Adaptation of DVI and external applications/services to vehicle status
   a) Vehicle enters safe operating condition (e.g., park gear, < 5 m.p.h., etc.)
   b) Driver adjusts vehicle controls (e.g., climate control, etc.)
   c) Suppression of hazard alert due to safe speed
Use Case 8: Adaptation of DVI and external applications/services to local regulations
   a) Application blocked
   b) Application suspended
   c) Interface modality disabled
   d) Age restriction
   e) Content restriction

For details on these use cases, download the FG Distraction Use Cases report.

Wednesday, April 3, 2013

Using smartphones to prevent traffic jams

Paul Leroux
Smartphones and driving don’t mix, right? Normally, you would get no argument from me. Driving is the one activity where a half-second lapse in attention can translate into a lifetime of regret.

But you know, there’s more than one way to use a smartphone in a car. Take Honda, for example. They’ve been experimenting with an approach that may help prevent accidents, rather than cause them.

Let’s rewind a bit. A few months ago, I wrote a post on the potential benefits of adaptive cruise control. These benefits include a dramatic reduction of traffic congestion and safer distances between cars.

Well, guess what: it seems that a smartphone app can have much the same effect. Recently, Honda equipped a number of drivers with an app that monitors acceleration and deceleration. When a subject drives in a way that avoids causing a traffic jam, the app screen turns green; otherwise, it shows blue.

Simple enough, right? And yet, the results were dramatic: formation of traffic jams was delayed by up 6 minutes and fuel efficiency shot up by 22%. Not bad for a smartphone app.

I am, of course, skipping a few details. Read more about the study in Tech-On!, an outlet of Nikkei Business Publications.

Jamming on a theme of connectivity at Automotive Megatrends 2013

Justin Moon
"The only time I really don’t feel connected is when I am driving in my car.”

You can always tell you’ve had a successful conference if you come away with a few “ah ha” or “oh no” moments. The above quote, which I cannot take credit for, was my “ah ha” moment at Automotive Megatrends 2013. The conference saw leaders and forward thinkers in the automotive industry come together and discuss (or debate) three technology streams: powertrain, safety, and the stream I participated in, connectivity.

The day began with a panel discussing the Big Picture of in-vehicle connectivity. Five panelists presented their views on business model pros and cons, where connectivity is headed, how it could change consumer expectations, and steps for ensuring success. Following the panelists’ presentations, the floor was opened for questions and dialogue with the audience. It was a great introduction and it set the stage for the rest of the day.

After a brief “networking” break, a panel discussing hardware and the effects associated with connectivity began. The format continued as before: each panelist spoke on their area of interest or expertise, then the audience joined the conversation. I was intrigued by the state of persistent storage in infotainment systems, including the problems and potential solutions for maintaining performance and reliability.

Lunch was just as engaging as the panels and discussions. I participated in a lively table-wide discussion with several analysts and industry thinkers on how the connection in my vehicle needs to become a seamless part of my lifestyle, just like my smartphone or the connected entertainment equipment in my home. This discussion was a great lead-in to the panel I had the pleasure of participating in — Software and Apps.

Clearing a path
The panel followed the same format as the others. Panelists discussed the role of the software platform and where application models fit into the future of the connected vehicle. One engaging panelist, after a rousing lunch discussion, explored how ubiquitous connectivity will clear a path for bigger possibilities with ecological initiatives, safety strategies, and making the vehicle a part of the connected lifestyle consumers already relate to and expect.

The final panel of the day was about wireless networks and had some industry big thinkers jam about infrastructure requirements, futures, and business models.

All in all, my experience at Automotive Megatrends was very positive and I look forward to doing it again next year.