On Tue, May 14, 2019 at 9:39 PM <[email protected]> wrote: > How would make your model walk. No demo needed for me? just simple talk > would be greatly > appreciated. >
You mean learn to walk, with robotic legs? Then set up an environment, where: - input = body sensors, translated into propositions - output = movement of each leg joint, also as propositions - reward = e.g. 10 if staying upright and moving forward, = -10 if fallen down Then just let the robot learn to walk on its own, using reinforcement learning. The transition function F would be updated via Bellman update as usual. Except that F has the structure of an RNN-within-RNN. (One RNN handles the proposition level, another RNN handles the predicate level). The time steps should be set very small for the AI to respond quickly. This is not the typical situation for a logic-based AI, but in theory it can work in this setting as well. I believe it is even practically feasible, just a bit unusual. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T3cad55ae5144b323-M59de8a9bb98be999cbae9a72 Delivery options: https://agi.topicbox.com/groups/agi/subscription
