We're not in immediate danger of any asteroids colliding with Earth -- at least not as far as anyone's aware. But it's not like it hasn't happened before, and there is an expected near-miss coming up in 2029. Accordingly, it's probably best to be prepared, and MIT researchers have developed a system that could help determine the best possible method to avoid a collision -- long before the situation becomes desperate.
An MIT team led by former MIT graduate student Sung Wook Paek describe a "decision map" in newly published research that would take into account the mass and relative momentum of an approaching asteroid, as well as the expected time we have before it enters into a so-called "keyhole" -- basically a gravitational halo around Earth that, once entered, all but guarantees the asteroid will collide with the planet.
The MIT-developed decision map basically details three different choices in terms of how to deflect an approaching asteroid: Launching a projectile at it to alter its course; sending a scout first to get accurate measurements to inform the best possible development of said projectile; and sending two scouts, in order to get measurements and also potentially nudge the object using propulsion, setting it up for an easier projectile-based knockout later on.
Time is the key factor in the model based on simulations run using asteroids Apophis and Bennu, two known objects we know relatively a lot about, including the locations of their gravitational keyholes in terms of proximity to Earth. The tests showed that with five or more years, the best course is to send two scouts and then the projectile. Between two and five years out, you're most likely to succeed with the single scout followed by a projectile fired from Earth. At one year or less, the bad news is that nothing seems all that likely to succeed.
The official plan for avoiding impacts from near-Earth objects involves potentially firing nuclear weapons at them, which is not a super-popular option. This method developed by MIT could help mean it never comes to that, provided our advanced detection methods are effective enough.