LidarhdBot

Bot application for running geospatial 3D mesh reconstruction algorithms on decentralised compute via discord messages.

Status

• Docker-compose-passing.
• Codepipeline-deploy.

Stack:

• Serverless messaging using: upstash.

• rendering visualization using: Streamlit.

• compute over Data using bacalhau:

• decentralised storage on top of[web3.storage]:

• bot service running on top of[discord.py]:

Credits

• bert lidar rendering pipeline:For the description concerning the geosptial data reconstruction algorithms.

• Dockerised discord bot template

• kafka examples

packages description:

1. Georender: Core package that implements the functions to convert the given pointcloud file to .laz 3D tile format.

2. Bots: Package for discord bot logic and commands.

3. aws_deployment: Scripts to deploy the necessary infrastructure on registered AWS cloud.

4. visualization: This is the rendering app written in streamlit that shows the result once the mesh reconstruction job is completed

Build instructions/Setup:

1. Locally running the bot:

1. We need to set following enviornment variables

   • For the kafka messaging service. the reference env file is defined in the .env.example here.
   • Then define the discord configuration in the config.json in 'bots/Discord/config.json', by following the steps defined here.

2. Then build the container application by running the Dockerfile for the streamlit and subsequentially that of the bacalau_sdk service.

2. On cloud instance (AWS):

1. Setup the awscli parameters for the account and access token.

2. Then run the scripts as defined below.

cd aws_deployment/ && cdk synth
# if its for the first time , you will be prompted to run the following command also
cdk bootstrap

• Follow the instructions in the readme of the aws_repository in order to provide the necessary permissions to your account in order to address errors.

4. Setup the personal web3.storage account and define the access token in the .env file.

5. Store the files on web3.storagefrom where you want to fetch the mapping format (shape/obj files). we have stored the reference files in data/ . and then keep the reference cid generated for the user as parameter to pass for generating the reconstruction job.

   • for more detials regarding the files latest version, checkout the following
     • https://geoservices.ign.fr/lidarhd
     • https://pcrs.ign.fr/version3

API's

discord bot api

/circum job_point <<Xcoordiante>> <<Ycoordinate>> <<ipfs_shp_file>> <<ipfs_pipeline_template_file>> : runs the reconstruction job for the given point coordinate and returns the job id.

/circum status <<job_id>>: checks status of the reconstruction job with given id.

/circum result <<job_id>>: fetches the result (laz file cid) once the job is completed.

/circum visualize <<ipfs_cid>>: generated laz file visualization link.

Detailled pipeline of the circum reconstruction process:

---
title: 3DP
---
flowchart LR
    user["`frontend with geo-coordinated tweet`"]
    confluentKafka["`submits the job for the bot`"]
    connector["that triggers the command when there is B/W"]
    botScript["`serverless script that
    - takes the input job request
    - calls the bacalau.py script to spin up new instance
    - passes the params coordinates to compute the resulting 3D file output
    `"]
    user-->confluentKafka
    confluentKafka-->connector
    connector--> botScript
    botScript --> job1
    botScript --> job2
    botScript --> ...
    botScript --> jobN

    job1 --> bacalau-publisher
    job2 --> bacalau-publisher
    jobN --> bacalau-publisher
    bacalau-publisher-->connector:Streaminng-results
    connector:Streaminng-results-->user

Loading