puklu/chronorobotics-project

Map Management System for Visual Teach and Repeat

Setting up the database

1. cd into project/.

2. run docker-compose up

Database should be up running on http://localhost:9090/. Login credentials can be set in the .env file present in the same directory.

a) Efficient management of maps

MinIO db is used as the database service. Once the service is up and running, the tool uploads maps to map bucket. The corresponding metadata for the maps for an environment is uploaded to env bucket in the db.

1. cd into project/scripts/Map_management/.

2. To upload a particular map for an environment from ~/.ros,

        python main.py -e <environment-name> -u <map-name> -snode <starting-node> -enode <end_node>
   

The tool can fetch maps from the db. Fetches the zipped map files from the db and then extracts them to ~/.ros/fetched_maps directory.

3. To fetch a particular map for an environment,

        python main.py -e <environment-name> -m <map-name>            
   

4. To fetch all the maps for an environment for the db,

        python main.py -e <environment-name>
   

5. To fetch only environment details (for eg map metadata). The details are saved as json in project/results/ directory.

        python main.py -oe <environment-name> 
   

Maps can also be deleted from the db.

6. To delete a specific map from an environment,

        python main.py -e <environment-name> -delamap <name-of-the-map>
   

7. To delete all the maps for a particular environment,

        python main.py -delmaps <name-of-the-environment>
   

b) Finding an optimum sequence of maps

Using the results of spectral anaylsis of the data in the db and metadata stored in env bucket, the mathematical model of the environment is used to plan a path between any two nodes of the environment optimising for a number of criteria.

1. cd into project/scripts/Map_management/.

2. To find an optimum path between two nodes,

        python main.py -e <environment-name> -findpath <first-node> <second-node>
   

The maps corresponding to the optimum path are fetched from the db into ~/.ros/fetched_maps/

Code from https://github.com/Zdeeno/Siamese-network-image-alignment.git is used for Neural Network related tasks for calculating likelihood of images.