How to implement probability / attraction for a patch in a random movement?

Situation: The agent moves according to Brownian motion. At each iteration, it can go in any direction, with the step ranging from 0 to a certain radius. Some patches have a feature that increases the likelihood of attracting the agent.

e.g. if there is a patch with the feature, the agent has a 70% chance of going to it and not to others.

Question: How do I implement this probability in the agent's movement?

Attempt: I managed to implement probability when I analyze only the agent's neighbor patches, using the move-to. But I need to implement this probability in a ray using fd.