Stellenticket Humboldt-Universität zu Berlin

Chem­nitz Uni­ver­sity of Tech­no­logy is a cos­mo­po­li­tan uni­ver­sity with strong regio­nal, natio­nal and inter­na­tio­nal net­works. It is home to about 2,300 aca­de­mic and admi­nis­tra­tive employees as well as more than 9,600 stu­dents from about 90 coun­tries. It is thus the third lar­gest uni­ver­sity in Sax­ony. In terms of the pro­por­tion of for­eign stu­dents, Chem­nitz Uni­ver­sity of Tech­no­logy occu­pies a top posi­tion among state uni­ver­si­ties nati­on­wide.

The rese­arch in the lab of Pro­fes­sor­ship Auto­ma­tic Con­trol and Sys­tem Dyna­mics focu­ses on deve­lop­ment of con­trol theo­re­tic methods in the fol­lo­wing areas:

• opti­mal and lear­ning-based con­trol for non­linear sys­tems,
• ana­ly­sis of uncer­tain, dyna­mi­cal sys­tems using set-based methods,
• hier­ar­chi­cal, opti­mal and fault-tole­rant con­trol,
• for­mal veri­fi­ca­tion of con­trol sys­tems.

The rese­arch in the lab has a strong focus on con­trol­ler deve­lop­ment for the full cycle star­ting from sys­tem model­ling to the end app­li­ca­tion. The par­ti­cu­lar topics also include advan­ced tech­ni­ques of data ana­ly­sis and clas­si­fi­ca­tion along with model uncer­tainty assess­ment. Spe­cial atten­tion is paid to com­plex and hier­ar­chi­cal sys­tems, where con­sis­tency bet­ween con­trol and sub­rou­ti­nes for dif­fe­rent sys­tem lay­ers is of par­ti­cu­lar inte­rest. Con­crete con­trol methods in turn include model pre­dic­tive con­trol, rein­force­ment lear­ning, adap­tive dyna­mic pro­gramming and other opti­mal con­trol methods, with a spe­cial focus on lear­ning aspects. Mathe­ma­ti­cal pro­ofs and ana­ly­sis of e.g. robust­ness pro­per­ties of con­trol methods in the pre­sence of uncer­tain­ties is a part of vir­tually every new con­trol method deve­lo­ped in the lab. We also try to bridge the gap bet­ween theory and com­pu­ter imple­men­ta­tion by expli­citly dealing with com­pu­ta­tio­nal uncer­tainty rela­ted to the imper­fec­tions of con­trol­ler imple­men­ta­tion in digi­tal and ana­log devices. For this sake, the lab works on a variety of mathe­ma­ti­cal methods which expli­citly incor­po­rate the said uncer­tainty and sub­se­quently aim at deve­lo­ping a basis of for­mally cor­rect and auto­ma­ted con­trol­ler extrac­tion.