Sharing our progress on implementing an (autonomous) differential drive micromobility vehicle in the Carla environment

Sharing our progress on implementing an (autonomous) differential drive micromobility vehicle in the Carla environment

(work in progress)

Project title: Scoomatic (autonomous scooter)

Who we are: research assistants and students from the chair of mechatronics at the university of Augsburg, Germany. Little experience in UE4.

Why we use Carla:

  • Complex environments already in existence
  • Open source
  • Variety of sensors implemented
  • “autonomous” pedestrians and vehicles available

Why we care about autonomous micromobility?
Simplified, a micromobility vehicle is every vehicle that fits on a bike lane. Most famous types are bikes, Segways self-balancing transporters and e-scooters. The most common shared micromobility vehicles are dock less e-scooter. Issues with nowadays dock less e-scooters are: they block sidewalks, need to be recharged manually and only get a sufficient coverage through oversupply. Therefore, some disadvantages could be eliminated if these vehicles could relocate their self autonomously.

Because we would like to use a differential drive vehicle, it is impossible to inherit anything from a wheeled vehicle. From our point of view, the documentation to add a new differential drive vehicle in Carla was a bit incomplete. Therefore, we inherited from a pawn and tried our best to mix the attributes of the existing pedestrians and wheeled vehicles.

Most changes to the source code can be reviewed in this repository https://github.com/ll7/carla under the scoomatic/fullreimp branch.

Modifications to the blueprints are harder to track, especially due to the fact that all assets are managed in a separate git-lfs repository. We haven’t found a “clean” solution, but we at least tracked the blueprints by modifying the .gitignore.

We currently update minor aspects of our documentation for the process of recreating our steps and the final documentation will be posted in the following repository: https://git.rz.uni-augsburg.de/luttkule/scoomatic-carla

Because our work is still under development and this post is anything but complete, feel free to contact us if you have any questions: lennart.luttkus[at]informatik.uni-augsburg.de

We would like to thank all contributors to Carla, and we are looking forward to share more of our progress in future updates.