Blog Archives

UBER Autonomous Car FAILED in 2018, but...given ANY other car, the same accident would have happened.

3/23/2018

Yes, this UBER car completely FAILED to even slow down, BUT... a human would have failed to slow down too, in fact, by the time a human could have possibly even seen this pedestrian, they could not get their foot on the brake in time, or swerve the wheel in time. Read on...

Let's be reasonable... there is liability on all sides enough to go around. No human could have stopped in time, there wasn't enough PERCEPTION TIME, or REACTION TIME, let alone braking time. Total distance needed to stop: 300 feet according to Caltrans. The pedestrian lady shows up on camera at 100 feet. This is also when a human could see her. 60 feet is just the perception time for a human at 40 mph. Also we know from several studies that humans can't see so well at night. In fact, in testing, humans 65 and older can only begin to see a pedestrian in the dark at about 65 feet. Young people 18-30 at 105 feet. This correlates with the video. You can only perceive what you can see.

Let's also not forget about the environment. It practically INVITED this lady to cross the street where she did, at about the most dangerous place you could jaywalk, she was on a sidewalk in the median that actually LED HER THERE. The median was decorated just like a park, and had a sidewalk that led to it from the bridge, but had no way out except to JWALK. This is a liability too. It is a pedestrian trap.

Illustration: Grant Johnson, TE

Bloomberg Forensic crash analysts who reviewed the video said a human driver could have responded more quickly to the situation. Really?

When I read this sensational article, I cringed at the bold statement. I too reviewed the video. The pedestrian was wearing DENIM pants and a BLACK shirt. Her bike was behind her and had no visible reflectors. I could see that there were only THREE lane stripes visible between the vehicle and the pedestrian, a distance of about 100 feet on Mill Ave. This would be the beginning of perception if a driver is alert. Reviewing the "Safe Stopping Sight Distance" standards from Caltrans Highway Design Manual Table 201.1 , it states that a vehicle traveling 40 mph (the speed of the UBER car) would need 300 feet of distance to safely stop. There was only had 100 feet. How is it then that these analysts are stating that a human could have responded quicker, when the traffic engineering knowledge in place says that more than three times the distance would be needed (300 feet needed to prevent collision) ?

Bloomberg Analyst states that UBER car should have detected pedestrian in median, and BRAKED. Really?

False. First of all, there is a little grove of trees and thick brush between the car's vantage point (below) and where the pedestrian/bike wouold have been in the median before entering the road. Is that how we humans drive, when we see a pedestrian near the edge of a roadway, we begin to brake? Hardly. We sail on past unless they enter the road. Why would programmers make autonomous vehicles brake whenever they saw a pedestrian that is not in the road? Its ridiculous to assume that, or that there would be a human behavior prediction model on random things. Also, there is no way software using lidar could effectively detect a human behind a grove of trees and bushes blocking the view, anymore than it could detect a pedestrian about to come around the corner of a building. It would ignore it just like a human would ignore it because they actually can't see it at all, in the dark.

STUDIES on VISIBILITY at NIGHT with DARK CLOTHES

The pedestrian hit by the UBER car was wearing a black shirt, and Levis.

Visibility of a Young Person: Age 18-30 = 105 feet
Visibility of an Old Person: Age 65+ = 65 feet

These are measurements while driver is looking slightly to the left of lane, and this visibility is also related to the fact that headlights are slightly aimed more to the right than they are to the left, a safety feature for not blinding oncoming traffic.

When a pedestrian gets in front of us, then is the time that we will brake, not when they are behind trees and bushes where we can not see them or prepare for their entry into the road, because they are hidden. These newspaper analysts are assigning blame to autonomous technology as if a human could have done a better job of it, as if they know, ...they don't know. The fact is, when pedestrians suddenly enter a high speed road into oncoming traffic, it takes a certain significant distance for a driver to first,
   1) PERCEIVE that they need to brake, then
   2) REACT by engaging their foot to the brake (or alternatively, swerving), and then the
    3) BRAKING time after the first two time periods to bring the vehicle to a stop, a time which depends on weather, tires, friction, weight of vehicle, etc etc.
The first step, in this case at 40 mph, takes 60 feet to perceive. The second step takes at least that long again, to react and move a foot to the brake, so the vehicle has effectively already moved 120 feet, at least, before the vehicle even begins to mechanically slow down by the braking process. The pedestrian is not even seen before 105 feet by a young driver, 65 feet by an old drive, so this pedestrian will be hit at FULL SPEED in ANY case with a human driver. You can't brake for that which you can not see. UBER had a chance to improve on this with LIDAR, but UBER did fail to deliver on shaving down that 120 perception and reaction time. On an autonomous vehicle this is theoretically supposed to be a near zero time, instantaneous computer processing, so they say. Was UBER software still trying to calculate what in the world it was seeing? If it knew, it could have saved this woman, like no other human possibly could. So I say, let's not bag on UBER and autonomous, let's be reasonable and hope for better autonomous reactions in highly confusing situations in the future.

Street Lights (luminaires) are in the Wrong Place. Lighting LIABILITY.

What appear to be pedestrian pathways are leading to darkened areas of the road at night. Why were these pathways built and not properly supported with signage and lighting? Why would they intersect the road at a location where drivers would least expect a pedestrian to cross (JWALK)?

I find it strange that there is a paved walkway / pathway in the median that leads to the street in two locations (the X path in the median), but the street lamps are positioned between these, effectively making the intersection of these pathways with the roadway completely in the dark at night. Coupled with trees and bushes, you have a serious visibility problem. This is the liability: Expectations.

There is a lot of LIABILITY to go around here, in Road Design, in Pedestrian and Bicycle Safety too.

Maybe it is just me, but this median is attractive. It actually looks like a park.
It looks like park with trees and pathways and rocks.
It looks inviting. But its in the wrong location. And at night, there is a serious problem with lighting in the wrong places, and pathways leading to a JWALK situation without a crosswalk or warning signs. At the ground level, it is confusing for both pedestrians and drivers who may see a ped crossing in a strange place, unexpected.

After taking a tour with Google street view, I could see that there is a sidewalk that is on the adjacent bridge that actually leads pedestrians into this DEAD END median "park" complete with pathways with no way out, except to JWALK.

See the pedestrians on the sidewalk that leads to the median which is just past the freeway ? There is even a pedestrian sitting booth in the side of the bridge here, a bridge with no way out up ahead. It is a beautiful walkway on the bridge, but with deadends at both ends of the bridge. Why are pedestrians using it?
If you build it, they will come.
Note: this bridge is one-way traffic for the SB traffic, Mill Ave.

A Dead-End sidewalk, with no safe way out, and no warnings for peds already in the median.

Sure, there is a sign there that says NO PEDESTRIANS and USE CROSSWALK (see photo below), but look at what direction it is facing. It is facing west, perpindicular to the sidewalk, to warn pedestrians on the other sidewalk to not come over into the median. In other words, if you are a pedestrian from the bridge on that side of this one way segment of road, you will not see this sign. You have no warnings on the bridge. In fact, if a pedestrian tried to view this sign they might interpret it to mean, "continue forward to the crosswalk" (in the direction of the arrow) rather than cross to leave the median. Then the pedestrian has this beautiful paved brick pathway that leads them, not to a crosswalk, but to a curb, faced with a decision to cross four lanes of high speed traffic without a crosswalk, without a warning, without a street light.

I can see many pedestrians making a serious mistake with this kind of design and lack of guidance. I believe there is serious liability here.

Architects, Planners and Engineers in their quest to make beautiful transportation facilities need to also think about safety in design and how the user will use the feature or facility. They must also think about if the transportation "art" is confusing, especially from a safety standpoint. This particular design should have had fences, no paved pathways to confuse, and certainly some guidance to peds. The pathways in no case should have connected directly to the curbs of the roadway as if to lead a pedestrian there. The bridge should not have encouraged pedestrians to use the median side of the bridges, as this can only lead to this dead end, on both bridges!

Also, a pedestrian seeing a paved pathway, a short cut, and a sign that says in effect "don't use me" is a very mixed message.

Grant Johnson, TE

Sources: All street view and aerial photos from Google Maps. Illustrations by Grant Johnson, TE

0 Comments

Autonomous Self-Driving Cars and Trucks and Transit

3/9/2018

1 Comment

Often depicted in way over simplified conceptual drawings, the Autonomous Vehicle is consistently misunderstood.

Autonomous Vehicles: In order for there to be compelling reason to incur the impact and expense of potentially retooling the entire vehicle industry, there would need to be compelling improvements to safety, capacity / efficiency, as well as the environment. The autonomous vehicle has in principle, the capacity to improve all three, but the conceptual illustrations one can find today on the internet when searching for "autonomous vehicles" leave much to be desired... they generate more questions than answers in one's mind. They are confusing. Such as why does a picture show all this "connectivity and sensors" from each vehicle, but the traffic pattern on the freeway is identical to existing conditions? Where is the benefit? Capacity increase? Or why would one think that it would be a good idea to show vehicles criss-crossing past each other at right angles in an uncontrolled intersection, narrowly missing bumpers, as if that is a safe idea or design? Logically, it's not. A roundabout would be a much better idea, lower speeds, and significantly reduced conflict points (goes from 9 potential conflicts down to just 1). In fact, it doesn't make any sense to introduce such a fallible and dangerous situation where serious injury can take place if technology fails in any way. Also, what about bikes and pedestrians in such a situation? Its as if there were no serious thought put into these concepts when it comes to having Complete Streets.

UBER Autonomous car hits pedestrian. news makes it look like autonomous is not safe

In my view, to be fair, a pedestrian was walking their bike across two lanes of a four lane boulevard that had a large median, in the dark, no crosswalks, becasue it was nowhere near an intersection opening (she was crossing a left turn bay first, an area where pedestrians are never expected. Technically, the pedestrian walking her bike was jaywalking in a high speed area, in the dark. In the video the bike is visible at a distance of about 60-100 feet (only three dashed stripes visible, 72 feet distance) . However, for a 45 mph roadway, the Safe Stopping Sight Distance is 360 feet. The car was traveling 40 mph, and a human would need 300 feet to stop in time. There is no way this was possible for a human driver to avoid even if they had seen her 200 feet back instead of 100. By the time a human's foot lifts from the floor to press the brake pedal, at least one second has gone by. They would have already hit this pedestrian before the vehicle even had a chance to slow down. Thne formula for this is dPRT = 1.47 Vt (US Customary), which means there is at least 60 feet of travel distance at 40 mph before the driver would even PERCEIVE that they need to hit the brake. Then there is the time for the foot to lift from the floor to the pedal and hit the brake. The pedestrian in the mean time, in this case, is hit, with or without an autonomous vehicle.

The cyclist/pedestrian did not look at the approaching car, until the last second, it was as if they were oblivious to the dangers of crossing a street, a straight street, where it would be extremely easy to see oncoming headlights. Who is at fault? In my view, absolutely the pedestrian. Because it was dark, this pedestrian was not readily visible with headlights. Yes, the UBER car's lidar or radar most certainly should have picked this up in the dark as it does not need light to function, but it failed on that. This is irrelevant as to why this accident happened, it would have happened with a regular driver and a regular car. A human would not have been able to react in time based on our most basic standards of road design. This is a completely unfair story and writeup, painting some narrative that driverless cars are more dangerous than human drivers. Not if this accident would have happened otherwise. How often pedestrians in the path of moving vehicles at night, have been hit. How dangerous it is for a pedestrian to cross a street in the dark, with oncoming traffic, and fail to yield the right of way, while remaining mostly invisible?

what components are in an autonomous car ?

Stopping Sight Distance = PERCEPTION TIME + REACTION TIME + BRAKING TIME
Autonomous vehicles theoretically change this equation to: SuperFast PERCEPTION TIME + SuperFast REACTION TIME + BRAKING TIME where the braking time remains the same because it is a function of tires, speed, friction, etc., but the Perception Time is a fraction of what humans need to make a decision that they need to brake...theoretically, and the Reaction Time is greatly shortened because no human foot has to move from the floor to get above the pedal to push it.

Any Autonomous Vehicle solution MUST also take into consideration the entire body of transportation modes, especially pedestrians and bikes. A pedestrian or cyclist will never be in the autonomous category, so these are mixed transportation use situations.

Good examples of extreme traffic situations can be found in China where density is consistently very high throughout urban cities. n Chongqing China there are residential and business skyscrapers that go on and on for miles and miles, averaging 30 stories tall.

Sidewalks are necessarily very wide, from 10 to 20 feet, to accommodate numerous pedestrians. The video to the right shows an area of the massive City of Chongqing where vehicle traffic has been completely separated from pedestrian traffic as a need. There are so many pedestrians in the area of skyscrapers, that it was not practical to have these interact with vehicles via crosswalks controlled by signal. There is not enough capacity.

The solution in this video was to literally BURY the traffic in a submerged roadway, and build a pedestrian square that extended numerous blocks and built on top of the submerged four lane road. The pedestrians never hear the traffic below, and safety is greatly enhanced.

THIS is a compelling reason to implement such an expensive change for the benefit of all, including drivers of vehicles. 90% of residents in Chongqing do NOT own or drive a car. As one watches the video of regular drivers on these massive 9-lane roadways, one can think of what benefit would come to the system if all vehicles were autonomous. In my view, safety would not only improve, but efficiency as well and capacity could be tripled as vehicles perfectly coordinate the merging, with tighter headways, and regulated speeds.

So What will an Autonomous Vehicle transportation system look like? What must it look like?

First of all, it must be safe for pedestrians. In the China video above, the pedestrians are completely separated from the vehicle traffic, because it is not safe or practical to ever have these meet in such high numbers. So there must be separation. There must be grade separation (or in the case of the China video above, a complete separation of vehicle traffic by tunnel if necessary). Grade separations are expensive, but to have a truly autonomous system and remove the human error factor that introduces accidents, often fatal accidents, a separation is needed. If a car is going 60 mph autonomously, and a child runs in front of it, the car no matter how automated, cannot stop in time, and fatality may occur, traffic will come to standstill, etc. We grade separate freeways, and now all roads with autonomous vehicles will need some method of separation in order to achieve the desired levels of safety, of capacity, of efficiency, even the environment and air quality. Fences. Ped and Bike bridges. Even signal systems tied in to the autonomous computer system, coordinated and optimized for safety of peds and bikes.

The future of traffic engineering will be to develop solutions that actually make sense, are safe and efficient, and which can take existing right-of way and turn it into a system where cars are separated from the pedestrians and bikes to improve safety, capacity, efficiency and air quality.