Successive Convexification in Learned World Models for Safe Offline Planning1 versionW_amucf24rv1 · current