Loading
0 Follower
0 Following
sa7show
Jianhang Sun

Organization

Huazhong University of Science and Technology

Location

CN

Badges

0
1
0

Activity

Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
Mon
Wed
Fri

Ratings Progression

Loading...

Challenge Categories

Loading...

Challenges Entered

Latest submissions

See All
failed 163656
graded 159300

Multi-Agent Reinforcement Learning on Trains

Latest submissions

No submissions made in this challenge.
Participant Rating
Participant Rating

Flatland 3

I have given the agent a sequence of move action, but it never appear on the map

Over 2 years ago

Sorry, I didn’t pay attention to this part.
So, maybe I can’t help with your debug process.

I have given the agent a sequence of move action, but it never appear on the map

Over 2 years ago

It seems that the probelem comes from flatland.

After I update the version of flatland from 3.0.2 to 3.0.5, this strange phenomenon didn’t happen during many times of testing.

Anyway, thank you.

I have given the agent a sequence of move action, but it never appear on the map

Over 2 years ago

I am not sure whether you can receive the notice of reply.
So I want to @you here Mr. @nimishsantosh

I have given the agent a sequence of move action, but it never appear on the map

Over 2 years ago

Thanks for you answering.

But I have checked the states of agents and found that when I give it move action its state is READY_TO_DEPART.

Here is picture showing an example of the problem I meet.

I print the triple of [position, state, action] of every agent at every step.
From the purple frame, we know that the initial_position of agent[4] is [21,8].
The red frame is the triples of agent[4].
The blue frame shows the first time I give it a MOVE_FORWARD(2) action and at this step its state is READY_TO_DEPART(1).
And the yellow frame shows that there is no other agent conflict with agent[4] since its initial_position is [21,8].

It is so strange that after I give it a move action, it didn’t depart but remained the state of READY_TO_DEPART(1).
After the blue frame, you can see that I give it more subsequent move actions, but it still didn’t depart but remained the state of READY_TO_DEPART(1).

This strange phenomenon does not happen every time, and I really can’t find the reason…

I have given the agent a sequence of move action, but it never appear on the map

Over 2 years ago


Here I gave this agent a sequence of actions, but it never appear on the map. And during this period, its initial position is free(no other agent in it).

I think even the actions are wrong, at least, it should appear on the map…
I don’t know why…

By the way, I set all speeds 1.0 and didn’t induct malfunction.

Flatland

Some questions ahout the agents' speed

Over 2 years ago

About the first question of this topic, @dipam has answered.

For the second question, my teammate emailed the administer and received the answer as follow:

"So rather than varying speeds it’s clearer to say multiple speeds.
There are different trains with different speeds. Their speeds are fixed throughout and never vary. Some trains are slower some are faster. Unlike round 1 where all were of the same speed.

Let me know if this clarifies your question."

Hope it is helpful to you as well.

Some questions ahout the agents' speed

Over 2 years ago

Thanks for your help.

Actually, I still feel puzzled about the β€œvarying speeds” at round2. Does it mean that the agents’ speeds are changeable every step during their traveling and we must check the agents’ speeds by function ’step’ every step?

The first part of this page makes me confused.

Could you please explain the β€œvarying speeds” at round2 in detail? Thanks a lot.

Some questions ahout the agents' speed

Over 2 years ago

I want to consult some questions about the speed of agent.

Firstly, I’m not sure about if I understand the speed correctly when it is smaller than 1. Assume that the speed of agent[X] is 0.7, and β€˜position_fraction’ is 0. After 2 steps of moving, does agent[X] reach the next cell and the β€˜position_fraction’ becomes 0.4?

Secondly, at round 2, the speed is varying. Does it mean that the environment can change the speed and we must check the agents’ speed by function ’step’ every step?

Is somewhere I can know more details about the speed? Thanks.

sa7show has not provided any information yet.