Intercept Guidance of Maneuvering Targets with Deep Reinforcement Learning

Algorithm 1

Guidance law learning algorithm based on TD3 algorithm.

Initialize critic networks , and actor network

with random parameters

Initialize target networks

Initialize replay buffer

for t=1 to T do:

  , reset the environment initial value

  While True:

   Select action with exploration noise, and

  observe reward and new state .

   Store transition experience tuple in

   Sample mini-batch of N transitions from

    ,

   Calculate the TD error

   Calculate the loss function

   Update the critic network using gradient descent as(where, e=1,2)

   if t mod d then

    Update by the deterministic policy gradient:

    Update target networks:

  end while

end for