1. The easiest way is to add an off-grid solar system to the trunk.  Many people have done this.  It requires no electrical modifications to the vehicle. There are many downsides.  
   1. A vehicle is a hostile environment for lithium batteries and inverters.  Adding a cooling system adds more cost and parasitic energy losses. 
   2. The added batteries add weight to the vehicle. Although 12 kg for a single LiFePO4 shouldn't matter much.  The additional energy on board should more than counteract the range loss.
   3. Battery degradation ruins the economics of solar charging. Even a cheap LiFePO4 that might last 4000 cycles adds $0.06 per kwh.
   4. The charging cable must be plugged into the vehicle's J1772 port.  Charging while driving is not permitted. 
   5. The overall efficiency of this setup is hurt by converting low voltage DC into 120v AC and then converting the 120v AC into 400v DC. Also, the MPPT charge controller is an additional conversion from 17v to 13v for the battery charging. 
2. Try to use existing vehicle hardware and hack it to work for solar charging.  The on-board charger is simply a large switch-mode power supply.  So in addition to AC power it should be able to charge from 100-300v DC. In a test, the Chevy Volt did charge from 120v DC. The charger reports RMS voltage over CAN bus. The rest of the car has no way of knowing that I fed it DC. 
   1. The vehicle's on-board charger typically has a maximum power of 3000-6000w. The efficiency of these chargers at 50-300w is likely to be low. 
   2. This method might also require plugging into the vehicle's J1772 port, preventing charging while driving. But maybe that interlock could be hacked around?
   3. The J1772 standard only allows current down to 6 amps. Maybe the voltage could be reduced to 100v and still charge.  An array capable of producing 600w won't easily fit on the roof of a car.  Efforts to reduce that charge current on the Chevy Volt were not successful. The charger output power is controlled by varying the output current.  The Volt's computer adjusts the output current in order to control the AC input current. It is possible to override the output current via CAN bus. But then charging will soon stop when the computer notices a large difference between the commanded current and the measured current.  This effort was abandoned due to concerns over point 2.1.
3. The best way is to use purpose-built hardware. This is the only method that would be considered by OEM manufacturers like Sono Motors or Lightyear. A dedicated solar charger should be slightly more efficient than an AC charger since the power factor correction circuitry can be eliminated. Additionally it will be sized for efficient operation at the array's actual output. 
   1. Connecting the new charger into an existing vehicle battery is its own can of worms. More on that in another update. 
   2. There is not a lot of suitable electronics on the market to do this.  Perhaps a microinverter could be hacked to function as the required DC to DC converter. Remember that microinverters have a complex anti-islanding algorithm that is necessary for safe grid connection. I felt that it was better to put effort into designing a custom converter.  The GZF inverter modules seem ideal at first glance, but those have no output voltage regulation. With the high open circuit voltage of a solar panel the converter can output 900v. No way is that thing going on my daily driver! Also, they are wound for a 12v input, not the 17v maximum power point of a "12v" solar panel. And they operate at a fixed voltage ratio, preventing maximum power point tracking.
4. A missed opportunity.  If I was writing the J1772 specification I would add a mode for "solar charge."  This would allow for cheap off-grid solar carports.   Instead of specifying a current limit, this PWM frequency would instruct the vehicle to use a maximum power point tracking algorithm.  The array would supply 100-300v DC that could be handled by the on-board charger. The rest of J1772 would work the same to ensure there is never live voltage on the connector when it's not plugged into a vehicle.   If these were possible businesses might like them because of no need to run electric lines, and no recurring electricity expenses.  They could be placed further from building so it would be less likely that people would just park there.  And customers would feel good about charging with solar energy. 

TL;DR

Making efficient use of the small solar array that can be mounted on a vehicle roof requires a purpose-built charge controller. There is little on the market meeting the requirements, so the charge controller needed to be built from scratch.

The J1772 AC charging port is not suitable for charging from a small solar array because it is designed to operate at 6 amps and above. If that could be overcome, the efficiency of a 3kw charger running at 10% power or less won't be that good. If it can't be overcome, it is necessary to add a storage battery which adds significant cost. Vehicle interlocks will prevent charging while driving.