Will Technology Bring an End to Wildlife Collisions?
Posted on September, 14th 2016
Having recently read an article on Smithsonian.com titled Will Driverless Cars Mean Less Roadkill, I decided the topic was worth further investigation.
Collisions with wildlife can be deadly. Your likelihood of encountering animals on the road are not unique to rural areas, and in fact are common in most suburban environments. One of the most common animals that pose a driving hazard are deer. Deer are most often on the move during dawn and dusk- two times of day when visibility can be compromised. The Geico Insurance website cites some statistics from the Insurance Information Institute:
- 6 million deer-vehicle collisions occur each year
- 200 resulting in fatalities
- More than $3.6 billion in vehicle damage
The highest risk states for deer crashes are speckled across the country, from the southeast to northwest. The astounding number of deer-vehicle collisions is a prime example of the tremendous impact wildlife collisions have on driver safety. Yet deer are not alone in the large animal-vehicle collision threat category. Depending on where you are in the United States, there could be frequent encounters with elk or moose as roadway hazards.
So, what happens when a driverless car encounters wildlife? The Google car can interpret larger animals as it does pedestrians, but it cannot currently identify small animals with its sensors, and therefore does not react to them. While car to car collisions may be eradicated by autonomous vehicles, they may not be able to outsmart animals, whose behavior is difficult to predict. Lasers, radar, and cameras are all in use to identify hazards in the roadway, but only large animals can be detected in the same manner that the vehicles detect pedestrians. Simply speaking, if the animal is large enough to be considered a large, soft object (as it interprets humans), the vehicle will react accordingly. Unfortunately, small animals will not be easy for autonomous vehicles to avoid- simply because the technology to detect and negotiate them as an obstacle to be avoided does not exist.
How does the driverless car handle large animals? The car either alerts the human occupant to assume control of the vehicle – or – the vehicle is programmed to respond itself. In the latter instance, the vehicle will make a decision based on what it believes the object in the roadway to be. In this scenario, the large animal will be interpreted in the same manner as a pedestrian, and the reaction would be to avoid colliding with it. As practical as this response sounds, it is not fool proof. Animals both large and small are unpredictable. The downside to the programmed response of the car is that the current position and speed of the animal is what the car will react to. If the animal has a change in speed or path, the car will not be able to adjust to those actions.
Predicting an animal’s next move is likely going to remain impossible. No matter how much observation occurs, there results are too inconclusive to adequately program predictive capacities into driverless car software.
What is a driverless car to do?
Probably nothing. We likely need to put the responsibility back to the human in the vehicle. In order to reduce wildlife collisions, we may need to look beyond self driving cars and instead look to other technology that can provide information to drivers, such as roadway sensors. Underground cable sensors have been shown to be able to pick up on objects moving as far as 10 feet away. Smarter roadways such as those equipped with this type of sensor system could provide drivers with alerts that there is a potential hazard in their path, thus signaling the driver to react. Or, our vehicles can be equipped with large animal collision avoidance systems like the one about to be unveiled by Volvo. This feature has a large animal detection with auto brake system that uses a radar sensor, front facing camera, and software to detect animals in or near the roadway, much like the autonomous vehicle’s pedestrian detection system. The difference is that we are still in a human driver situation with the ability of the car to override the driver actions- the vehicle uses its camera recognition technology, analysis, and processing with an image database- to get the car to decide whether to activate the system. This is the first car that combines large animal detection with auto braking.
If you are interested in more research about animal collision avoidance, please check the Western Transportation Institute’s Road Ecology Program