The agent only observes the first-person view in pixels, a compass angle, and a dictionary of inventory items [1]. So any solution would need to have a signific

The agent only observes the first-person view in pixels, a compass angle, and a dictionary of inventory items [1]. So any solution would need to have a significant computer vision component. [1] http://minerl.io/docs/tutorials/first_agent.html#taking-actions http://minerl.io/docs/tutorials/first_agent.html#taking-acti...