Required Operator Training for Autonomous Vehicles?

We have talked here in the past about the difficulties autonomous vehicle drivers (operators?) will have acclimating when control of their vehicles is handed back to them, such as when they approach a work zone. Studies with simulators have shown a need for anywhere from 4 to 14 seconds for a driver to get a full grasp of all of the relevant external factors they must consider as they begin to drive.

A recent article in Axios Autonomous Vehicles points out that aviation has made use of automation for some time now. And they, too, understand the problem of moving from automated to human operators. In aviation, training focuses on that hand-off. Pilots are drilled in flight simulators on a variety of potential problems. So, when they encounter that problem during a real flight, muscle memory takes over and they react quickly and correctly.

The recent 737 Max 8 crash further underlines the importance of that training. It was apparently not included and that may have contributed to the pilot’s difficulty in regaining control.

The difference between aviation and autonomous vehicles is that training is mandatory for all pilots. If you fly a 767 you must stay current in all 767 training. However, for vehicles, a big selling point is that drivers no longer have to drive. They are told they can act more as passengers – gazing out the window, catching up on work, or watching an endless variety of streaming entertainment. Getting from that idea to one of mandatory training is a very long stretch!

Adding to the problem are the very different ways automakers are designing the machine-to-human hand-off. Each one is different.

In the Axios article, they quote Steve Casner of NASA, “We’re terrible at paying attention — and we think we’re awesome at it” Mr. Casner argues that drivers will need training. And they will need continuous updates to that training in order to learn how to deal with automation. Without initial user training and frequent refresh classes drivers will quickly become complacent.

This is a new topic of discussion but one that we must have to make CAVs safe for work zones and other segments of roadway with changing conditions.

New Open Source AV Visualization Tools May Aid Our Industry

One of the challenges the roadway safety infrastructure industry faces in regard to autonomous vehicles is understanding how those vehicles visualize the world they are passing through. Manufacturers have been restrained in their sharing of that information. The best we can get out of them is “Keep doing what you are doing to make striping, signing and traffic control devices easier to see.”

But a story published yesterday in The Verge by Andrew Hawkins details efforts by GM Cruise and Uber to make some of those visualization tools open source and free to use. It is even provided in a fairly simple and easy to use format that anyone can use on most any device.

 

This could be very useful for pavement marking manufacturers or contractors. It may be helpful for sign manufacturers. And it will definitely help traffic control device manufacturers understand what the vehicle “sees” and what it does not.

Now this is far from the ultimate testing platform, but it will help our industry begin to develop an understanding of the underlying issues and ways we may be able to address them. It may also help work zone ITS providers in that it offers a simple data formatting system that may be able to accommodate data feeds from smart work zones.

The GM Cruise tool is called “Worldview” and can be found HERE.

The Uber tool is called “Autonomous Visualization System” or AVS for short and can be found HERE.

We haven’t spoken with anyone who has used these tools yet. So, please try them out and tell us what you think. Are they useful to our industry? And, if so, how? What can be improved? We look forward to hearing from you!

 

Drivers Recognize the Importance of Connected Work Zones

We have been talking for the past couple of years about “connected work zones” – that is, the automatic and real-time method of putting our work zones on the digital map that everyone is quickly coming to depend upon when choosing a route.

We have argued that traffic control workers don’t need more to do when they are setting up or tearing down a work zone. So, to arrive at a point where we have timely and accurate reporting of work zones, it must happen automatically.

Several companies are now providing solutions. Those solutions vary in their complexity and technologies involved. But in their simplest form they each include a device attached to existing traffic control devices. One of those is normally the arrow board. The beauty of this approach is that when the arrow board is turned on, the system immediately tells the digital map that a work zone just popped up on that route at that precise location. And when it is turned off, it tells the map that the work zone is now gone. It happens every time a “smart” arrow board is used and those are becoming more and more common.

We all “get” this. But now the driving public is also recognizing the importance of these systems. An article by Tim Harlow in the January 27th Minneapolis Star-Tribune talks about a system supplied by Street Smart Rentals to Minnesota DOT in the Twin Cities.

He points out that the existing 511 system does a good job of informing the public about long-term projects, but that short-term and unplanned closures can cause just as much disruption yet are not included in their warnings to the public.

The system supplied by Mike Granger and Street Smart Rentals is changing that for the better. And with the arrival of autonomous vehicles, this will become even more important. In the article Brian Kary, MnDOT’s Director of Traffic Operations “said the technology is not active now, but it could be this summer or fall. MnDOT is evaluating costs before making it a permanent 511 feature. The agency also is setting up a timeline install the technology and figuring out how best to get information to other traffic information sources, such as Google, Waze and TomTom, since not everybody uses 511.”

We believe economies of scale will quickly and significantly reduce those costs. And the need for this information will bring down any barriers to those traffic information sources. We look forward to hearing more about this system and others like it the exciting year to come.

How Does the Traffic Message Channel Work?

We’ve talked in the past about the need to update the work zone information on digital maps in real time. But how does that process actually work? The answer is surprisingly simple while offering far more detail than you might expect. It is sent over FM radio and satellite channels using RDS-TMC protocols. RDS stands for “radio data system”. TMC stands for “traffic message channel”.

The information is sent in very small packages several times a second within a frequency used for digital identification of the station, song titles, etc. In this way, location codes and event codes are sent without interrupting the audio and updates any navigation devices in very near real time. That information can then be used in calculating the fastest route. It will also recalculate as incidents occur that cause significant delays.

In the United States the digitally coded traffic updates are distributed by Navteq over FM channels and by Sirius/XM satellite radio. iHeartMedia and TeleAtlas also provide commercial services in about 77 US metro areas.

Once received, the codes are automatically displayed in the driver’s preferred language making them more readily understandable and therefore more effective.

Each incident is digitally coded and sent as a TMC message. Each message consists of an event code, location code, expected incident duration, and other pertinent details. The message includes one or more phrases describing the problem. The first portion states the problem and the second portion gives clarification regarding the types of vehicles affected, recommended actions by the motorist, etc.

As you might expect, there are many work zone related messages. In fact, there are more than 150 work zone specific messages as well as many hundreds of messages just focused on queue length, travel delays, and lane closures. The work zone messages get fairly specific: mentioning pavement marking, resurfacing, bridge work, water main work, etc. They even mention temporary signals in one.

There are also many messages about incidents, weather, and special events.

The RDS-TMC system was developed before wide-spread use of GPS. So, they do not use a lat and long to identify the location. Instead location is described in relation to major intersections and points of interest.

As work zone reporting becomes more sophisticated, codes can still be added to provide additional detail such as the lanes that are closed, the length of the closure, expected delays, and more. Only a little more than half of the code capacity has been used so far. So there is plenty of room to grow. And that is important. Because autonomous vehicles will require far more detail. Discussions are already underway regarding what new details must be included and the formatting, etc. for them.

A New Approach to Geolocation

In our last post we discussed an interesting discussion on automated vehicles and work zones that took place as part of ATSSA’s Midyear meetings. Another automated vehicle presentation was made during ATSSA’s Sign Committee meeting. Mr. Jamie Retterath of Vergence Automation (https://vergenceautomation.com/ ) began by discussing the relative advantages and disadvantages of different kinds of sensors. He pointed out that no single device works best in all conditions. He suggested a combination of sensors and software is the best way to “see” in all conditions. The car’s software would then choose the sensor image with the most contrast and clarity.

But the most interesting part of his presentation was a way of positioning vehicles anywhere they travel and in any weather conditions. GPS is not accurate enough to guide vehicles by itself. And the geometry of the road changes frequently due to construction, variations in pavement markings, etc.

He called them fiducial signs, meaning points of reference. Even with a foot of snow, these signs would tell vehicles exactly where they were in relation to the road. A series of small signs, perhaps as small as 12” square, would be posted on both sides of the road similar to what’s shown in the photo above.

These signs would be posted in the digital map. Sensors would see the signs and triangulate their position from them. Anytime the road geometry was changed the next vehicles that drove that stretch would recognize the change and would send that data in to change the digital map. It is a simple and relatively inexpensive way to speed the adoption of autonomous vehicles.

We asked about work zones. Mr. Retterath said the first autonomous car to come across a closed lane would see the obstruction and drive around it.  Like other changes it encounters, the vehicle would report the closed lane and it would update the digital map in near real time. The same could be done for lane shifts, crossovers, or other geometric changes.

These signs would improve location accuracy and could help speed the adoption of autonomous vehicles, thus saving many lives. Its just a concept at this point, but it represents yet another clever way of moving Towards Zero Deaths.

 

Automated Vehicle Roundtable Held at ATSSA Midyear Meetings

The American Traffic Safety Services Association (ATSSA) recently concluded their annual Midyear meetings in Williamsburg, Virginia. Their Innovation Council met on August 23rd. But before the meeting began officially, they held a joint round-table discussion with members of the Automotive Safety Council. The ASC represents manufacturers of automotive safety system components including cameras, LIDAR, radar and other sensors.

The ASC led off by presenting a sort of Automotive Sensors 101 class that explained the different technologies, what they do well, and what they don’t do so well. This was a big help to ATSSA members who must design traffic control devices that these sensors will be able to “see” and react to in the very near future.

Cameras used for lane tracking look out about 500 feet on highways with a viewing angle of 40 to as much as 100 degrees. The viewing distance decreases on city streets while the viewing angle increases. As camera technology improves, they plan to hold the lane keeping range to 150m as there is little benefit to extending it. Instead they will widen the field of view to better detect pedestrians, balls rolling into the street, etc.

Cameras currently see black & white (gray scale) and red. White lane markings are much easier for cameras to see than yellow because white has far better contrast.

The ASC maintained (as we do) that digital maps must be updated in real time. Long term work zones are easy enough to include in digital maps. Short term work zones are more of a problem. And chip seals are the worst as they are short term AND include no pavement markings – just chip seal markers.

As we move from level 3 to 4 and 5 automotive system hardware won’t change much. It will probably decrease in price, but that’s all. Rather the system functionality and human-machine language will be the key differentiators. The algorithms used by the vehicle to decide what is important, what is not, and how the vehicle should react will constantly evolve and improve.

The ASC shared their market forecast for growth in the next few years. In 2020 the first level 5 vehicles will be sold. Level 2 (driver assist) vehicles will total about 13 million vehicles. By 2030 more than 90 million vehicles will have at least level 2 automation and level 5 will total nearly 3 million vehicles. But that means less than 5% of all vehicles on the road in 2030 will be level 5.

There are still very different approaches to level 3 automation. At level 3, vehicles will automatically center in their lanes, follow a route and stop when required. But unexpected conditions, such as work zones, causes the vehicle to return control to the driver. Some manufacturers see level 3 as a step toward levels 4 and 5. But others, especially Google, feel level 3 is dangerous and so will not produce cars requiring human control at any time. Level 3 peaks in 2025 at 2 million vehicles then drops as level 4 and 5 vehicles become more popular and available.

The ASC group told us control will be ceded in work zones. But how that will happen is not clear. Still, they agreed with us on the need for sufficient time for the driver to acclimate before having to make important decisions.

Once the ASC concluded their presentation, Scott McCanna of David Evans & Associates made a presentation from our industry perspective and asked several thought-provoking questions about work zones along the way.

When channelizing devices including cones, drums and delineators are used to redefine a lane, will device spacing become important for automated vehicles? Will we need to maintain some minimal spacing to hold CAVs attention? And what happens when one or two cones are knocked down? Will the automated vehicle become disoriented? Or revert to the old lane markings?

It was further suggested than CAV logic should see drums and cones as a higher priority when choosing a direction of travel than existing pavement markings. Drums, cones, etc. should indicate a change…perhaps one that automatically triggers driver control in the case of Level 3 CAVs.

The time went by very quickly and everyone agreed it was a great first step in building better understanding between our two industries. Future meetings are already planned to build on this and plan for our future.

 

National Dialogue on Highway Automation

Being the work zone data nerds that we are, we attended the National Dialogue on Highway Automation Workshop #2: Digital Infrastructure and Data held August 1st and 2nd in Seattle. The first workshop covered planning and policy. Workshop #3 focuses on freight. #4 is Operations and is held at the same time as the National Rural ITS meeting in Phoenix. The final workshop will be held late this year in Austin and will be more technical in nature as it covers infrastructure design and safety.

Each workshop includes a series of presentations followed by breakout groups where ideas are discussed and then shared with the larger group. The format works well and benefits from the input of a wide range of stakeholders.

You will be happy to hear that work zones came up early and often. In fact the opening comments used work zones as an example of the need for some sort of standardization as every agency now provides varying amounts of data, different types of data, different formats and a very wide range of detail. Another speaker called work zones the “low hanging fruit” for highway automation in general and data collection and dissemination in particular.

There were about 200 in attendance and maybe 30 raised their hands when asked who attended the Automated Vehicle Symposium last month in San Francisco. So, this was an almost entirely new group.

You should also know the FHWA is seriously committed to this process. They had 20 or 30 of their own people at this event running it, moderating the breakout sessions, and asking lots of questions.

There were a number of themes that jumped out at us. One was data quality and verification. The consensus was that state DOTs will probably have the responsibility of verifying data accuracy. But what that process might be is unclear. It will likely vary by data type. In our case it will probably come as a quality check after it is already posted. Work zone activity must be reported in real time to be actionable, so they will weed the inaccurate reports (and reporters) out after the fact.

Remarkably most in the room were well acquainted with the MUTCD. Multiple comments suggested that it needs to be revised to recognize automated vehicles. Some even suggested reducing the leeway states have in specifying sign formats, pavement marking details, etc. to create more consistent traffic control for CAVs. But later others pointed out this is unlikely to happen and the effort would be better spent doing this outside the MUTCD process, at least to begin with.

These two days were time well spent. If you are able, we strongly encourage you to participate in one of their future workshops, especially the event in Phoenix. It will be focused on traffic operations. But because it will be held in conjunction with the NRITS show, it will also spend more time on automated vehicles and rural roads.  Learn more HERE.