Intersections have been a constant source of problems for drivers since the first two Model Ts crashed into each other. We can consider the set of challenges facing autonomous vehicles trying to navigate their way through traffic more complicated when they have to deal with the unpredictable behaviors of human drivers.
1. Intersection Complications
The A.I. of the vehicle must perform the same maneuvers as their human counterparts when encountering an intersection, such as stopping correctly at the wait-line designating the crosswalk, choosing the correct path through the intersection, and perform these maneuvers while determining the current human behavior scenario. Are there pedestrians in the street and if they have crossed against the light.
Of course, there are many variations to this scenario that the A.I. must take into consideration to prevent human or car accidents in intersections.
Human behavior can be a random thing. John Smith knows the red light law when crossing the street, but feels the yellow light gives him the authority to run fast through the crosswalk. The autonomous vehicle at the wait-line cannot drive through the intersection when the light turns green because Smith is still illegally running through the crosswalk.
Visibility can also be a complication for the automated vehicle. The A.I. needs a simple line of sight to make its determinations. An illegally parked car too close to the intersection could be partially across the wait-line and blocking sight of on-coming traffic for the car.
2. How They Decide
Autonomous cars use sensors to determine how fast other vehicles are traveling. Using this information, the system calculates the predicted pattern of travel for the vehicle. This pattern also weighs the probabilities of factors that also influence that pattern such as driver distractions, vehicle mishap, or speed changes.
Modeled predictions are constantly being updated as the self-driving vehicle moves through the intersection. New data and risks assessments are being processed to determine if and where obstacles or collisions might appear.
Autonomous vehicles also must deal with unpredictable elements like pedestrians. The A.I. examines other variations than just walking pedestrians. Pedestrians may carry heavy loads, be looking at cell phones, or talking to another pedestrian.
The pedestrian may pay attention to a car out of sight of the autonomous vehicle and altering their walk pattern abruptly. Walk gait may even change for no observable reason beyond the whim of the walker.
Autonomous vehicles take in observed data like the direction a pedestrian is looking, the influence of other pedestrians in the crosswalk, or the speed and color of the walk signals to mitigate the risk of endangering a pedestrian.
Cities will need to adapt for the coming day when autonomous vehicles outnumber manned vehicles. Architects and developers will have numerous forms of self-driven transport to deal with besides the car. Many new challenges will develop from delivery units to Uber’s air-taxi. Without rebuilding from the ground up, adapting current cities will be difficult in concept and execution.
Eliminating private vehicle parking will allow the expansion of streets, freeing intersections of view obstruction. Building flexible sections of city blocks for drop-off and pickup by delivery and self-driven vehicles is important to freeing up precious parking real estate.
Allowing designated autonomous crosswalks for smaller delivery vehicles of fast-food or packages would also clear traffic from streets. These would be like bike lanes, but just for the delivery vehicles of a smaller nature.
These redesigns would assist in intersection navigation by self-driven vehicles because of much more predictable patterns of behavior and monitoring of other autonomous vehicles on the streets.