A couple of thoughts here ...
This system would rely on pretty much universal adoption in order to work effectively (and somehow would have to be retrofitted to older vehicles), so the cost of this is going to be quite significant. And, if it slows the development and introduction of self-driving technology, the benefits may be questionable - whereas autonomous vehicles with self-driving features could be introduced gradually.
In addition, although multi-vehicle accidents are common, most accidents are either single-vehicle or involve pedestrians/cyclists (Australian Bureau of Stats figures here), in which circumstances the technology would likely not be very effective.
Still, if it works and can be effectively implemented, who am I to criticise?