The key point in all this is 5G (vs LoRa(WAN)), as autonomous vehicles is one of the major driving (please excuse the pun!) use cases it has been designed and rolled out for. It has a cost and you will need to pay to play…'cause:
“it has been designed and rolled out for”,
doing it on the cheap using a ‘roll your own’ solution vs one that has been developed, engineered and ‘right sized’ for the target applications - by a global industry, ecosystem and standards body(ies) is unlikely to work (well) and likely a futile experiment for a solo developer or even a single company. Not to discourage your enthusiasm and dreams, just want to set realistic expectations here.
The L-A has not particularly architected LoRaWAN implementations for the task you have defined so far, whether <Ghz (with potentially up to 64 channels available in your local territory if not in many of the other places you might call out), or in 2.4Ghz band (though potentially globally deployable, standard spec currently limits to just 3 channels as noted above - with also sorts of real world implementation issues, conflict with other 2.4Ghz solutions not withstanding - WiFi, BT/BTLE, Zigbee, proprietary or whatever…), rather I have seen suggestions from L-A participants and contributors that LoRaWAN might play a part in the autonimous vehicle market as an ancilliary and back up solution, e.g. 5G network goes down perhaps can still (geo)asset locate (note NOT track realtime) any disconnected or stalled assets (vehicles), side channel reporting of potential higher latency/lower priority data such that in aggregate it helps offload the main critical path high data rate/low latency network services etc. (Yes, in doing so, help protect the ‘expensive’ bit as you have identified! 
)
Whilst your initial higher refresh rate/data rate/payload content can be ‘value engineered’ down to more manageable levels I’m sure - perhaps using logical analysis and thought experiemnts such as those suggested by Nick @descartes above, you really need to consider hard what is the absolute minimum you can get away with in your use case, and even then consider the consequences of non-delivery of the associated messages for many reasons - some outlined above and range issues such as highlighted by Stuart @LoRaTracker. Remember another reason why cellular is often more more expensive is that it is on licensed bands and in many territories those licences come at a cost - often £$€ billions! - which is over and above the costs of hardware and infrastructure…(BTW just saw Reuters reporting that US Cellular has sold some of its licences to AT&T for $1Billion just today!)
Of course one of the advantages of such environments is that the RF environment is cleaner and better controlled vs the noisy/dirty ISM bands making for potentially more reliable (but still not guaranteed!) delivery, allowing for targeting data/timing critical use cases such as oooh. let me think, I know : autonomous vehicles! 
