Apparently this happens already on some single seaters: when body roll starts, the roll centre moves from its normal position a short distance above the road, half-way between the wheels, towards the outside of the corner. I have never seen any attempt to use this phenomenon to control body roll.
I
did an analysis of my Morris Minor's front suspension. As standard, the roll centre is about 3"/75mm above the road, and migrates in a small circle through the range of roll, only having deviated 2"/50mm at a rather extreme 6deg of roll. Then I considered replacing the lower control arms with leading arms which, geometrically, are swinging control arms of infinite length. Now, the roll centre migrates rapidly in the direction of roll, i.e. towards the outside of the curve, so that at a mere 3deg of roll the roll centre is outside the track.
It should be obvious that this represents a self-correcting system. If there were zero roll stiffness, the vehicle would adopt that angle of roll at which the resultant of weight and lateral inertia passes through the roll centre, and roll no further. Then there is no roll moment, as the moment arm length is zero.
In the example of my Morris this translated to a requirement of 0.67g of lateral acceleration to achieve 1deg of roll. The exact ratio is variable, and depends on the length and angle of the upper control arms. In the example roll stiffness is discounted and all weight transfer derives from forces effectively acting on the roll centre. There are nevertheless many ways in which this system may be tuned, e.g. varying front and rear roll centre heights; varying the rates of rc migration front and rear; using the system only at one end of the vehicle, and some other system at the other; or introducing a tiny amount of roll stiffness at one end only.
The effect of the last-named is to introduce a very small roll moment. Even so, the system will generate only a tiny fraction of the torsional loads associated with conventional suspension systems tuned by large roll-stiffness differentials. It is therefore another way to get a car to handle decently without the need for a torsionally-stiff vehicle structure - which is an ongoing obsession of mine.
The use of leading or trailing lower control arms also suits the system to mechanically interconnected suspension, a la Citroen 2CV, which I've always thought a brilliant piece of work. The roll-correcting characteristic allows one to achieve resonant frequencies in the 1.0-1.2Hz range without the 2CV's tendency to scrape its door handles on the road without any of the wheels losing contact...