I am asked to design an algorithm to find the shortest path in a given map with obstacles. It is actually not hard with negative reward and using value iteration to solve this. My problem is, if consider the episodic MDP (5 minutes), each robot can only take in the map, and the robot can never reach the goal within 5 steps, what should we do now? Can we optimize getting the coffee in that time limit? Would the same reward (say -1 in each block) apply to the map?

Optimize the path for coffee!!

  • $\begingroup$ @AlexR. I am confused by the saying that "we need a policy that optimizes getting the coffee in that time limit", what does episode actually mean here? Do we need the immediate reward in the non-reachable areas? $\endgroup$ – xxx222 Feb 15 '17 at 0:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.