Argo and Tekton are projects that come to mind. Also, if you try to implement this, you will see that encapsulating your pipeline in a single file gets disorganized quickly.
This is not looking to be the case right now.. Companies are doing quite well without explainable models. SDC companies like Waymo are hoping it will be sufficient to:
1. Produce a low failure rate estimate, based on millions of virtual miles/fake cities.
2. Combine several NN outputs in a logical way. That way there are at least some interpretable values that lead to the result. e.g. a separate model that produces candidate trajectories of surrounding vehicles, one to perceive sensor input, one to plan, etc.
Forgot to mention if you listen to accounts of blacking out while freediving, it's clearly much different than a physical concussion. People wake up as if they just had a deep nap, with generally no symptoms as the day goes on.
Blacking out while freediving is much different than the type of "blacking out" you normally would associate with sport injuries.
It's an evolved response the body does to conserve oxygen use in the brain. Consciousness shuts off, but the brain is still active. For example, someone who blacks out this way will not get water in their lungs--a part of the brain that is still active keeps your mouth shut so that no water is accidentally inhaled.
Freedivers try to avoid blackouts, but in this context, there are established safety procedures and it is not nearly as bad as you may think.