Skip to content

Latest commit

 

History

History
159 lines (83 loc) · 5.34 KB

README.md

File metadata and controls

159 lines (83 loc) · 5.34 KB
Raw

Instructions

Phase 0: system update

Pull from the upstream remote

git pull upstream
git merge upstream/montreal21

Make sure you have an updated system using

dts desktop update

Update Duckiebot

If you have a Duckiebot, also do

dts duckiebot update

Phase 1: Walkthrough of notebooks

Run

dts exercises build
dts exercises lab

Open the web link that will appear. You are in Jupyter Lab in the folder solution. Open the notebook braitenberg01.

Work through the notebooks in sequence.

The notebooks guide you to fill out some functions in preprocessing.py and connections.py.

They will also indicate the use of other tools.

Phase 2: Experiment in finding connections, modifying the agent

In this phase you have to experiment with the rest of the agent.

The skeleton is in solution/agent.py. Read through and see how it uses the things you defined in the other Python files.

Feel free to change anything.

Do note that you will most probably need to edit the BraitenbergAgentConfig class! Its current gain and const values are off. They are much larger than they should be, as a way to help you finetune them.

At the begining of an evaluation episode, the max and min values for both motors will be off, but as the agent lives, it'll adjust itself. This slight change in motor values is normal.

💻 Testing in simulation

To test in simulation, use the command

$ dts exercises test --sim

There will be two URLs popping up to open in your browser: one is the direct view of the experiment (probably http://localhost:8090). The other is VNC and not useful for this exercise. Ignore it.

This simulation test will be very slow! We suggest opening the simulation viewer and enjoying a cup of tea/coffee while your agent does its job. Monitor how it acts! You might get some ideas on how to fix your matrices or the agent.py.

This simulation test is just that, a test. Don't trust it fully. If you want a more accurate metric of performance, continue reading to the Do local evaluations section below.

🚙 Testing on the robot

You can test your agent on the robot using:

dts exercises test --duckiebot_name YOUR_DUCKIEBOT

If you remember the video lecture, this is the modality "everything runs on the robot".

You can also test using

dts exercises test --duckiebot_name YOUR_DUCKIEBOT  --local

This is the modality "drivers running on the robot, agent runs on the laptop."

Phase 3: Make a submission

At the end, to submit your homework, you should submit your agent using

dts challenges submit

The robot should travel at least an average of 2 meters from the starting point.

(Note: we might change the target to make it easier if we see it is too hard in the following days. Please understand that we are still calibrating against the huge variety of participants.)

If an error of this form occurs

Traceback (most recent call last):
  File "/usr/local/lib/python3.8/dist-packages/duckietown_challenges_cli/cli.py", line 76, in dt_challenges_cli_main
    dt_challenges_cli_main_(args=args, sections=sections, main_cmd="challenges")
  File "/usr/local/lib/python3.8/dist-packages/duckietown_challenges_cli/cli.py", line 203, in dt_challenges_cli_main_
    f(rest, environment)
  File "/usr/local/lib/python3.8/dist-packages/duckietown_challenges_cli/cli_submit.py", line 165, in dt_challenges_cli_submit
    br = submission_build(
  File "/usr/local/lib/python3.8/dist-packages/duckietown_challenges_cli/cmd_submit_build.py", line 41, in submission_build
    raise ZException(msg, available=list(credentials))
zuper_commons.types.exceptions.ZException: Credentials for registry docker.io not available
available:

you need to log into docker using dts. Use this command:

dts challenges config --docker-username <USERNAME> --docker-password <PASSWORD>

Happy life all together in harmony

We run the cloud evaluation service based on limited resources from our universities and sponsorships. To avoid a denial-of-service-by-many-exercises-submissions attack, we implemented a priority queue: the more submissions you send, the lower priority you have and it will take more time to do them. This is fair in the sense that, if you have never submitted, you are on top of the queue. However, we try to evaluate all submissions.

We hope this works and we don't need more stringent policies.

You can help us in the following ways.

Retire obsolete submissions

Note that you can "retire" submissions.

If you do this:

dts challenges submit --retire-same-label

then the old submission to the challenge will be related.

There is also a command dts challenges retire.

Finally, you can do this on the Challenges website.

Do local evaluations

We suggest you evaluate your work using the command

dts challenges evaluate

This should take a few minutes. This is not supposed to be an interactive process: just let it run, and when you return, you will find the output in a folder, including videos, and trajectories, and all the statistics you find on the website.