At a red light, a queue grows backward even though every vehicle is traveling forward toward the intersection. When green begins, the queue does not vanish at once: discharge at the front and new arrivals at the back jointly determine how the queue boundary moves.
Six signal cycles unfold beside a time-space diagram. Random arrivals produce an observed queue, while a second shape comes from a traffic-wave calculation using arrival and discharge rates. Their comparison shows what a compact analytical model can predict and where individual vehicle timing adds variation. The signal splits and following behavior make the cycle-by-cycle growth and dissipation visible. This matters for estimating storage needs and spillback risk. A queue's length is a dynamic consequence of both demand and service, not a number that can be inferred from the red duration alone.
Four green lights along a corridor do not guarantee a smooth trip. If their timing is unrelated to travel between intersections, a vehicle released by one signal may arrive at the next just as it turns red. The same road can therefore produce repeated stops without any change in demand or geometry.
Watch film
An intersection contains several traffic streams that would like to move at the same time. The problem of signal phase design is to allow compatible streams together while separating movements whose paths cross. The video turns that spatial safety rule into a graph problem.
Watch film
The center of a conventional intersection forces many turning paths to compete for the same space. A knotted intersection rearranges those paths and schedules passage through several smaller conflict areas, allowing movements to weave through a compact geometry in a controlled rhythm.
Watch film