Skip to main content

Project Enceladus Landing page

Landing page for all material related to Enceladus, developed by Rhea team
Created on February 14|Last edited on September 1

Context

This short project has taken inspiration from the Woven Planet Level 5 blog on Improving ML Models for Autonomous Vehicles, which outlined their use of W&B Tables to help them understand safety critical corner cases such as pedestrian detection.

Overview

Project Enceladus has been initiated to establish a new Autonomous Vehicle Agent for our organization. Due to the large scope of the overall project, it has been split into a number of self-contained projects, each of which tackle a single functionality of the AV Agent, such as Semantic Segmentation, Depth Estimation, Lane Detection, etc. Each self-contained subproject under Project Enceladus is supposed to run in parallel during the first three quarters of of 2022 and will be split into a number of sprints. This would allow us to work through the last quarter of 2022 on putting all the separate functionalities developed under Project Enceladus together. This would enable us to have a working Autonomous Vehicle Agent by the end of 2022.
In the first sprint of Project Enceladus we try to achieve the following:
  • Establish a new Semantic Segmentation Architecture
  • Establish an iterative and explainable workflow for improving our Semantic Segmentation Model.
  • Ideate of additional self-contained projects for functionalities such as Depth-estimation.


Project level stats


Runs
100
Artifacts
294

Business context and sponsors

Project Scope

Team members

Slack Integration: Enceladus channel

Github repo

Technical Dashboard

Current project metrics and evolution over time

Run set
313


Status Reports



Technical Notes / Reports

Notes for researchers, including a deep dive into our methodology and more!

project