One of basic approaches, as you have suggested, can be based on Euclidean distance between the two vehicles. This can be achieved using Vector Field Histogram block.
You can refer to the following example for more detail about implementation:
There are other algorithms specifically for multi agent collision avoidance such as: Velocity Obstacles, Reciprocal Velocity obstacles etc.
Also, if you just want to plan paths for multiple agents from start to goal locations there are different algorithms such as: Conflict based search, M* etc.
But they are quite complex to implement. For simple tasks the first approach should work fine.
