To recap, the most recent flight-tests failed because the airframe ~ broadly configured as a cube with a propeller at each of its eight corners ~ was arranged with 'X-8' logic, in so far as all eight motors were involved in control motions. At its heart the problem lay in the fact that only the upper drone was constructed in alloy whilst the booth and lower drone were made of plastic extrusion throughout. Without the upper end tested independently, we have to conjecture that the three-dimensional space-frame in its entirety was insufficiently rigid enough to resist the resonance and flexure that play havoc with computerised control.
In order to resume the test program it will be necessary to fly the upper drone whilst attached to a vacant accommodation booth underpinned by an identical undercarriage of the same outline. That done the lower drone can be tested independently in its role as merely providing sufficient thrust to offset the bulk of the weight of the payload. The approach has much to commend it as a first step, as motors run at constant RPM use considerably less energy than those required constantly to vary it in accordance with the demands of the flight controller.
Fortunately it would appear that the airframe of each drone may be left substantially unmodified, as the picture above shows. Each of the perimeter extrusions measures around 30" but including the tube-connectors the outside edge of each square is 32". In turn this means that if the motor-mounts are offset upon the airframe then there is sufficient clearance for the propellers. A quick check using classroom Pythagoras confirms that a measurement of 16" along the shorter sides of the triangle bound by the hypotenuse (in green) provides for around a half-inch of spacing between rotors.
At the same time we'll take the oppo to reconfigure with the more aerodynamically efficient tic-tac-toe arrangement of rotors, whose outlines on the upper drone appear in red whilst those on the lower appear in blue. You might ask why not arrange them otherwise, but the suggested arrangement allows for regular ingress and egress for the passenger at the bottom end, whilst at the top or steering end we revisit a config that has been successfully flown in the past... likely because it is more rigid than the alternative of controlling motion using a motor at each corner.
What emerges from all of this is effectively an abandonment of the notion that upper and lower drones might be identical in order to reduce the part count to a minimum. In design, as in life, there is ever a compromise between the perfect and the practical.
