Collisions are at the origin of catastrophic disruptions in the asteroid Main Belt and play an important role in the production of objects that are injected into unstable zones and transported to Earth-crossing zones where they are characterized as Near-Earth Objects (NEOs). Collisions are witnessed by the observation of asteroid families, each composed of asteroids which originated from a single parent body, broken-up by a collision with another one. Understanding the collisional process and its outcome is required to study the collisional evolutions of small bodies or planetary formation. In the last five years, for the first time we have successfully performed numerical simulations of high-speed collisions, which also account for the production of gravitationally reaccumulated bodies. Our first simulations using monolithic parent bodies have succeeded in reproducing the properties of some well-identified asteroid families, showing that gravitational re-accumulations following disruptive collisions are the key process accounting for the existence of asteroid families. Then, we have investigated the effect of the internal structure of the parent body on the outcome properties. We found that the most likely internal structure of large asteroids in the main belt is not monolithic but rather composed of macroscopic fractures and voids.

I will briefly review our current knowledge on the origin of NEOs and their source regions, and then make a review of the physical concepts and simulations of collisional disruption. NEOs are certainly fragments of larger asteroids of the Main Belt, injected either directly or by diffusion into main resonances that transported them to Earth-crossing orbits. Our simulations suggest that most NEOs with diameter larger than several hundreds of meters should then correspond to highly shattered bodies or gravitational aggregates. I will then show that the recent images of the 500 meter-size NEO Itokawa provided by the first sample return japanese mission Hayabusa are consistent with this result.