Markus Hinsche

We released Tensorflow training code which tries to implement the paper: https://github.com/merantix/imitation-learning

Hi everyone, we have been working on source code for the training. In my opinion, the information for this is scattered across multiple repos (+ papers): - https://github.com/carla-simulator/imitation-learning has inference...

@Johnnyswong It is not stopping in the training data recorded, therefore it is not what the model learned to do.

Here is another code that also does the training, which might or might not be easier to read :) https://github.com/merantix/imitation-learning

Praneeta! In Tensorflow, the method `dataset.map()` has a parameter `num_parallel_calls`. See how we use it in our training implementation of this paper: `https://github.com/merantix/imitation-learning/blob/master/imitation/input_fn.py#L100`

During a training run on the cluster, I also only see the logs once (not duplicated), i.e. so far I only observed the behavior on my local setup.

I tried adding this to src/common/depthmap_toolkit/toolkit.py and src/common/depthmap_toolkit/visualization.py. However when running `python src/common/depthmap_toolkit/toolkit.py src/common/depthmap_toolkit/tests/huawei_p40pro src/common/depthmap_toolkit/tests/huawei_p40pro/camera_calibration.txt`, I still see duplicate logs. ``` python src/common/depthmap_toolkit/toolkit.py src/common/depthmap_toolkit/tests/huawei_p40pro src/common/depthmap_toolkit/tests/huawei_p40pro/camera_calibration.txt DEPRECATION WARNING: The system version...

We need unit tests for `src/common/rgbd_toolkit/`

Yes, maybe due to a slightly different implementation, the parameters outlined in the paper do not make our car drive well. If I remember the steering wasn't learned well which...

> 1.You didn't implement lateral augmentation in your code,right?By lateral augmentation,I mean 3 cameras transformation.If yes,is it done in the given data (24GB) ? We didn't implement this augmentation but...