Today, I was challenged with the question of how well robocars would deal with deer crossing the road. There are 1.5 million collisions with deer in the USA every year, resulting in 200 deaths of people and of course many more deer. Many of the human injuries and crashes have come from trying to swerve to avoid the deer, and skidding instead during the panic.
At present there is no general purpose computer vision system that can just arbitrarily identify things — which is to say you can’t show it a camera view of anything and ask, “what is that?” CV is much better at looking for specific things, and a CV system that can determine if something is a deer is probably something we’re close to being able to make. However, I made a list of a number of the techniques that robots might have to do a better job of avoiding collisions with animals, and started investigating thoughts on one more, the “flying bumper” which I will detail below.
Spotting and avoiding the deer
- There are great techniques for spotting animal eyes using infrared light bouncing off the retinas. If you’ve seen a cheap flash photo with the “red eye” effect you know about this. An IR camera with a flash of IR light turns out to be great at spotting eyes and figureing out if they are looking at you, especially in darkness.
- A large number of deer collisions do take place at dusk or at night, both because deer move at these times and humans see badly during them. LIDAR works superbly in darkness, and can see 100m or more. On dry pavement, a car can come to a full stop from 80mph in 100m, if it reacts instantly. The robocar won’t identify a deer on the road instantly but it will do so quickly, and can thus brake to be quite slow by the time it travels 100m.
- Google’s full-map technique means the robocar will already have a complete LIDAR map of the road and terrain — every fencepost, every bush, every tree — and of course, the road. If there’s something big in the LIDAR scan at the side of the road that was not there before, the robocar will know it. If it’s moving and more detailed analysis with a zoom camera is done, the mystery object at the side of the road can be identified quickly. (Radar will also be able to tell if it’s a parked or disabled vehicle.)
- They are expensive today, but in time deep infrared cameras which show temperature will become cheap and appear in robocars. Useful for spotting pedestrians and tailpipes, they will also do a superb job on animals, even animals hiding behind bushes, particularly in the dark and cool times of deer mating season.
- Having spotted the deer, the robocar will never panic, the way humans often do.
- The robocar will know its physics well, and unlike the human, can probably plot a safe course around the deer that has no risk of skidding. If the ground is slick with leaves or rain, it will already have been going more slowly. The robocar can have a perfect understanding of the timings involved with swerving into the oncoming traffic lane if it is clear. The car can calculate the right speed (possibly even speeding up) where there will be room to safely swerve.
- If the oncoming traffic lane is not clear, but the oncoming car is also a robocar, it can talk to that car both to warn it and to make sure both cars have safe room to swerve into the oncoming lane.
- Areas with major deer problems put up laser sensors along the sides of the road, which detect if an animal crosses the beam and flash lights. A robocar could get data from such sensors to get more advanced warning of animal risks areas.
Getting the deer to move
There might be some options to get the deer to get out of the way. Deer sometimes freeze, a “deer in the headlights.” A robocar, however, does not need to have visible headlights! It may have them on for the comfort of the passengers who want to see where they are going and would find it spooky driving in the dark guided by invisible laser light, but those comfort lights can be turned off or dimmed during the deer encounter, something a human driver can’t do. This might help the deer to move.
Further, the car could emit loud sounds. There should be study of what sounds are best at scaring deer — though we don’t want to scare other deer onto the road. There might also be other light patterns that scare, rather than transfix a deer.
An extreme step would be to arm the car with a plastic BB gun (sometimes called airsoft.) Ok, you’re thinking, are we going to arm robots already? But a plastic BB in the rump might be just what is needed to get the deer to jump away. Clearly we need some good safety around such a weapon, and might even insist that the human passenger pull the trigger.
A research project to learn what combinations of lights, sounds and even plastic BBs do the best job of reliably scaring deer off the road sounds like a good idea. Reportedly, while they sell devices that make an ultrasonic noise, they do not work according to insurance reports. I would presume something that sounds like a cat or other predator would be a good goal. Electric vehicles will be making minimal noise.
Hitting the deer
If all the above fails, the car may hit the deer. In particular if the deer suddenly jumps out onto the road from a place where it could not be seen, you might still hit it even at a good speed. (If it’s further down the road, the impact would be slow.)
The deer weighs as much as 300lb. Moose and some other animals can be even bigger. Smaller deer will damage the car but should not cause an accident for a robot driver. But there can be lots of damage. And in some cases, deer fly up and go through the windshield, resulting in major risks for the people inside.
One immediate thought is that robocars don’t need to have windshields at all, though passengers will like them for the view and to eliminate motion sickness. However, they don’t have to be big wide windshields with no obstruction. They could be made out of panes with strong bars between them. Alternately, a grid of metal bars could come up in front of the windshield when traveling in deer country. Or if they can be raised quickly, the bars could snap up just before any crash. (This could be added to a human driven vehicle as well.) Compressed air can accomplish such a quick deployment. The robocar will know at least a second in advance in most collisions.
There is of course the problem of what passengers do during a sudden stop or worse, an impact. In a multi-person robocar, half the passengers will be facing backwards in most cases, and they will fare better. The stop will be quite disturbing to the forward facing passengers. The use of seatbelts would be strongly recommended while moving at high speed through deer country at the busy crossing times. If necessary, the car will always be monitoring the passengers in such situations and firing airbags to protect them in case of impact.
In a non-impact situation at lower speeds, the robocar would of course reduce the braking and even give audible warning before doing it.
Another approach, valuable in any sort of collision, is what I will call a “flying bumper.” This is a bumper which is mounted on two long pistons. A blast of compressed air is able to send the bumper out so it moves 2 meters in front of (or behind) the car. The bumper is then able to be pushed back into its pistons, resisted either by compressed gas, springs or even metal forms being destroyed. This adds 2 meters of travel in which to accomplish a large fraction of the total deceleration of the impact. After the bumper returns to the car, it then presses against the car’s regular crumple zones, which absorb as much of the impact energy as they can.
The extra 2 meters (or possibly 3 or 4 on longer cars) could mean a huge difference in the amount of shock the passengers (and car, and deer) are subject to. A life and death difference even with high speed collisions with solid objects, let alone deer.
If the collision is to be with another car equipped with a flying bumper, then a total of 4 meters of extra travel could take place. Suddenly even a head-on collision or brick wall collision is much more survivable.
Such a system could go on a human driven car. The trick is to have a computer system which is able to reliably detect an impending accident and to work out precisely when impact will occur. That can be done even if a human is at the wheel, and technologies that do variants of this, like Mercedes Pre-Safe, are already on the market. But key to this approach is the knowledge that even if a robocar might get into a collision, it will very rarely get into one it doesn’t know is coming well in advance. At least well in advance for a computer.
This bumper can also have airbags in it. Airbags not for the passengers but for the person, animal or even other car being hit. Again, knowing the moment of impact allows the computer to fire the airbag before the impact and soften the blow. This is particularly important if it’s a pedestrian being hit. Airbags might also solve the windshield problem, in that an airbag that explodes in front of the windshield just before the deer hits it might disperse the force over the whole windshield, and soften the impact. Airbags are currently expensive because of the safety constraints of the job they perform.
The bumper could also be set so that one side comes out further than the other, creating a diagonal bumper which directs the deer away from the car instead of trying to throw it in the direction of travel. The robot should be able to compensate for those forces.
The flying bumper seems very valuable to me, though I do see one problem with it. We must work very hard to assure it never triggers at the wrong time, particularly if a car is stopped. You want a fail-safe that can prevent that. One way to do that might be to only charge the compressed air tank that pushes out the bumper with a pump driven from the drivetrain while in motion, and dischrage that tank when stopping. That’s a bit wasteful of energy, but there might be a way to have a fail-safe system that controls where the compressed air can go based on movement.
Today a computer can’t, like a human, just look out into the world and tell what everything is, including a deer by the side of the road. However, the above techniques suggest it doesn’t need to. (Some day it will be able to, adding even more ability.) In many ways I believe the robocar should actually do much better at detecting and avoiding deer, particularly in the dark.