The risk of having an accident in a car could soon be drastically cut thanks to a new approach to wireless networks.
The sensors promise to help negotiate dangerous traffic
Embedded sensor networks are designed to replace existing networks that can only share information that has already been captured and stored.
Sensor networks instead take information from several individual nodes - potentially thousands of them in different locations - in real time, and can act on it accordingly.
This means that during a journey, vehicles could monitor each other's speed and position - and therefore dramatically cut the risk of accidents.
"I think that we may, in the future, go beyond just communication to using the network to interact with the environment," Professor PR Kumar of the Convergence Lab at the University of Illinois told BBC World Service's Digital Planet programme.
"For example, cars on a highway may talk to each other and find out each others' speeds. So a sensor in that context could just be the speedometer on a car, which could talk to surrounding cars.
"These cars could then co-operate with each other to avoid accidents and so on."
Most advanced networks currently work on an ad hoc basis - established through links between laptops, for example, which can share information between them.
But embedded sensor networks represent a shift away from computers communicating purely over a network, to sensors which communicate with each other.
And the next stage is an "actuator network" - where computers are able to act on the information they receive from the sensors. This could mean, for example, reducing speed in advance if slower traffic conditions are detected.
"What we really want to do is interact with the physical world, consisting of the cars," Professor Kumar said.
Heathrow's forthcoming driverless pods work using sensors
"Interaction is a two-way process, so I don't just want to know what speed it is; I want to change the speed.
"This gives rise to sensor-actuated networks, which are deployed over a wireless network. That could be the kind of system we are headed towards."
He said that these systems represent "the convergence of communication, computation, sensing and actuation."
However he conceded that a lot of issues remain to be sorted out before such a network could be put into practice on the roads.
One issue is that of who controls the cars - whether it is a centralised computer system or whether the cars should just collaborate with each other.
"That's a question of how you make decisions, whether in a centralised way or a distributed way," he said.
And there are also issues regarding the priorities of the networks, and how they may conflict; basic functionality - for example ensuring the cars do not crash - will have to be built in regardless of other functionality.
"This gives rise to interesting design challenges," Professor Kumar said.
"There are all kinds of challenges that we do have to confront.
"But things have been getting complicated without our knowing it. For example, a typical car on the road may already have 60 or 70 microprocessors. So we already have taken a step into this world of complexity."