I'm interested in Steering Behaviors by Craig W. Reynolds in a project but my agents are in a discrete grid (think Minecraft). Essentially all agents internally have continuous speed, accelerations and force vectors (first and second derivate vectors) but their positions are constrained to a grid and updated in fixed timesteps. How would I adapt a framework like this one in order for the agents to be on a grid on different cases:

  1. The first and second continuous derivates are shared with the agent, positions are discrete.
  2. The derivates are NOT shared, and have to be estimated
  3. The action mappings are also required to be discrete (forward/back, left/right, rotate 90 ect)
  • \$\begingroup\$ How fast do your agents move in grid terms? \$\endgroup\$ Commented Apr 1, 2023 at 13:53
  • \$\begingroup\$ Let say the grid G has spacings of one meter and agent has an internal real speed of 5.2 m/s and the game world is updated every half second. Then on each update the agent would move 2.6 meters and would snap to the nearest coordinate in G. Then from an external viewer the estimated speed would be 4 - 6 m/s on G. \$\endgroup\$ Commented Apr 2, 2023 at 2:19


You must log in to answer this question.

Browse other questions tagged .