Applications of H-infinity controller design as presented in the literature often rely significantly
on manual tuning and iteration. This paper proposes uncertainty lumping, a novel framework
which can increase the automation of the tuning process, simplify the weight design and reduce
the conservativeness of the controller synthesis. Uncertainty lumping combines unstructured and
parametric uncertainties while preserving their dependency relationships. This enables arbitrary
combinations of competing and dependent requirements, along with system uncertainties, to be
incorporated into the synthesis process. Furthermore, worst-case stability requirements can now
be expressed as weights. To showcase the potential of this framework, the Thrust Vector Control
(TVC) of a rigid-body launch vehicle is considered. A gain-scheduled robust controller is designed
for the ascent phase. The applicable requirements from the VEGA rocket are translated into
simple, well-known weights and lumped together into an equivalent weight that encodes the robust
stability condition, as well as the stability requirements for nominal and worst-case conditions. The
resulting controller is successfully validated against the ESA-i4GNC non-linear simulator and by
frequency analysis, where the lumped weights are employed. The H-infinity norm of the synthesized
controller is compared to that of the textbook approach, demonstrating reduced conservativeness.