Sorry about that I gave you wrong information. I mixed up VMware Fusion on macOS capability with that of Workstation/Player on Windows/Linux. On Fusion, a Windows VM would have a virtual USB camera and the camera of the Mac hardware didn't have to be connected to the VM. I have a laptop with Intel RealSense but I haven't tried connecting it to any VM for facial recognition. It appears as multiple devices on the Windows 10 host Device Manager but appears as just a single Intel composite device in the VMware USB connect menu. Anyway, you could try connecting webcam to the VM but I would think you would need to install the 3D/IR drivers of the camera into both the Ubuntu host and the Windows VM for the facial detection. But I suspect if it appears as just a single device in the VMware USB connect menu, it may not work at all. I haven't spent any effort to try to get the facial recognition of Intel RealSense to work on a Windows VM.
As for the DisplayLink USB dock, it looks like DisplayLink has driver for Ubuntu 16.04
http://www.displaylink.com/downloads/ubuntu
My workaround to avoid the freezes with the Windows 7 laptop several years ago with the DisplayLink adapters was to restart the laptop during my lunch break at 12pm otherwise inevitably it freezes sometime between 12:30pm and 1:30pm; but even then this was not foolproof as sometimes it still freezes around 3:30-4:00pm despite having the mid-workday reboot. Back then, I get the impression DisplayLink does not seem motivated to fix the problem (assuming it is some sort of DisplayLink driver issue) as their business model seem to be to sell to the OEM. So if the larger chunk of their revenue/profit is attributed to the OEMs, consumer complaints about the freezing may not get proper treatment unless the OEMs complain. I haven't used the DisplayLink adapters for a while now (stopped using them early 2014) so I have no idea with regards to the stability of their drivers now to use them nonstop for a full work day.
You may also want to look at this
Not sure if you can take this KB an encouraging sign. But in 2012/2013, DisplayLink didn't seem interested at all in fixing the freezing problems.
As for using ESXi, you would need at least two physical (or virtual) machines. One to host the ESXi and the other machine to access the VM(s) running in the ESXi. Of course you can create an ESXi as a VM and still have more VMs running inside this ESXi VM and you could have this setup in all just one laptop (this is when the virtualize Intel VT-x/EPT is needed). Running ESXi probably isn't the solution as it just add another layer of complexity for you and if you want good video capability inside the VM, ESXi can't provide that without GPU passthrough but that is another topic in of itself.
