[ODE] Combining (low-level) physical simulation and (high-level)
AI/user control (long text :-)
Remi.Coulom at univ-lille3.fr
Mon Feb 14 16:12:19 MST 2005
Tyler Streeter wrote:
>I'm not sure if using genetic algorithms for training/optimization is
>the best method for more complex behaviors. Lately I've been
>researching other learning techniques (e.g. temporal difference
>learning, see Reinforcement Learning by Sutton & Barto) that seem more
>biologically plausible (see papers by Wolfram Schultz on reward
>signals in the brain).
You may find my PhD thesis useful:
I managed to use reinforcement learning to teach a 2D dynamical system
to swim. Interactive demos (for Windows) are provided on the link above.
I am now trying to connect my reinforcement learning algorithms to ODE.
More information about the ODE