At the recent IEEE Conference on Intelligent Transport Systems, MIT professor Berthold Horn presented an algorithm for alleviating traffic flow instabilities. According to Horn, who is employed at MITs Department of Electrical Engineering and Computer Science, such instabilities arise from sudden brake manoeuvres in dense but otherwise flowing traffic. If one driver brakes, the following driver has to brake even stronger to avoid a collision, and thus the velocity variations build up upstream along a lane of traffic. According to Horn, this effect characterizes a chaotic system where positive feedback generates velocity oscillations. Even Adaptive Cruise Control systems (ACC) cannot avoid the occurrence of such instabilities.
Horn conceived a model that describes such an instable, oscillating traffic system. It is based on a damped-wave equation. And he found an algorithm that could stabilize such behaviour -in terms of maths it can be described as a Lyapunov function. Variables in this function are driver's reaction times, their desired speed and their eagerness to reach that travelling speed which in turn translates into the speed at which they accelerate as soon as they see gaps in front of them.
The algorithm could be implemented in ACC systems. However, to be effective they will have to be modified: While today's ACCs only measure and control the distance to the vehicle ahead, a system based on Horn's algorithm would also have to take into account the distance to the following car. Which means that the vehicles sensor landscape would have to be extended by an additional backwards-looking radar or lidar sensor. Or, since Horn is expert in computer vision, by a camera-based, backward-looking system.
His system however has a downside: It only functions if a high percentage of vehicles is equipped with it. AAC systems however, still are used predominately to premium class vehicles - radar and lidar sensors still are too expensive to deploy them in more affordable vehicles. To overcome this chicken-and-egg