“…[23,47,39]. Robust methods [4,44,11,32,1] rely on Huber [57] or Vapnik losses, leading to support vector regression [41,50,55] for state space models, and take advantage of interior point optimization methods [67,81,113]. Domain constraints are important for most applications, including camera tracking, fault diagnosis, chemical processes, vision-based systems, target tracking, biomedical systems, robotics, and navigation [53,98].…”