deepdrive-zero
deepdrive-zero copied to clipboard
deepdrive
Top down 2D self-driving car simulator built for running experiments in minutes, not weeks
We currently use the bike model to calculate the center (x,y), and angle of the vehicle using slip angle. However, when we rotate the four points defining the top-down ego...
Log stats to the arcade window when visualizing an agent with perhaps graphs and/or percentiles of things like g-force, jerk, speed, etc...
We currently use an MLP for the actor and critic networks in PPO. This _should_ be fine so long as the environment is fully observable, but presents scaling issues when...
We currently allow moving the steering wheel and accelerator across their entire range in one frame which is not realistic. More here https://github.com/deepdrive/deepdrive-2d/blob/d4d8fcc40842cffb7473663e1181394ac947061d/deepdrive_2d/envs/agent.py#L1048-L1050
We currently have a lot of TODO's in the code around using Numba to accelerate various things like denormalizing actions, g-force calculations, and physics stepping, but I'm not really sure...
Bumps [torch](https://github.com/pytorch/pytorch) from 1.3.1 to 2.2.0. Release notes Sourced from torch's releases. PyTorch 2.2: FlashAttention-v2, AOTInductor PyTorch 2.2 Release Notes Highlights Backwards Incompatible Changes Deprecations New Features Improvements Bug fixes...
![dependabot[bot] avatar](https://avatars.githubusercontent.com/in/29110?v=4 "dependabot[bot] avatar"){kind=avatar}