tl;dr: I think ego-road is what we want, not ego-vehicle and wish to ask for some feedback. Any comments/questions/thoughts can save me from being trapped in "ego" rabbit hole.
Tangential to damping concept from CP1.3, I was wondering whether ego-vehicle is a precise concept. The damping in braking modeling term $\text{damping in braking} = \min\left(1, \frac{s_{follow}^*}{s_{follow}}\right)$ reflects both _following driver's braking behavior_ or _lead driver's tendency_ to move away. However, for inferential purpose, having one modeling component represent two physical components can cause inferential degeneracy and not recommended. Partial improvement may be "ego-road (interval between two vehicles)" rather than "ego-vehicle". This decreases one degree of freedom (from N cars to N-1 intervals) which is a better model that prevents degeneracy problem.
After all, wasn't road's summary statistics (e.g. relations between flow, density, speed) been our main interest, not nitty gritty details of "sense of self-esteem or self-importance (definition of ego)" of vehicle's driver?
| | ego-vehicle (agent: $vehicle$) | ego-road (agent: $vehicle_{n}-vehicle_{n+1}$) |
| ------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| interpretation of damping in breaking | $vehicle_{n}$ behaviors that affects as crash prevention:<br>1. signaling to warn $vehicle_{n+1}$ to prevent $vehicle_{n}-vehicle_{n+1}$ crash (i.e. hit from the back) <br>**and**<br>2. carefulness to prevent $vehicle_{n-1}-vehicle_{n}$crash (i.e. hit the front)<br>**and** ... | $vehicle_{n}-vehicle_{n+1}$crash is prevented by {signal from $vehicle_{n}$ **or** careful $vehicle_{n+1}$ **or**...} |
| cartoon (agent:brain interacts with environment:earth) | ![[Pasted image 20240229113129.png]] | ![[Pasted image 20240229113116.png]] |
| justification for modeling | there are thousands of behavior that affects crash prevention (hence **and** from row1) and we may need to justify why certain ego-vehicle's behavior is ignorable. | we don't really care what behaviors (hence **or** from row1) inside environment affects crash prevention. they are all black-boxed as one variance term in stochastic model |
| control (measured by feeback loop clockspeed) | $vehicle_{n}$ has more control on 2 than 1 which may be needed to be modeled with stochastic model's variance term | road infrastructure designer like government may have different control over roads e.g. road in BrasÃlia is easier to control than road in amazon forest (if there's any) |