The ethical tradition that grounds moral obligation in duties binding on any rational agent, rather than in the consequences an action produces or the character it expresses. Its central claim is that some things may not be done to a person even when doing them would produce more good overall.
Kant gave it systematic form. The categorical imperative in its first formulation tests a proposed maxim by asking whether it could be willed as a universal law without contradiction; in its second, better-known formulation it requires treating humanity, in oneself and others, always as an end and never merely as a means. The move from that to modern rights language is short, and the concept of human dignity in postwar constitutions and declarations is substantially Kantian in ancestry — a person is not a quantity of welfare to be traded off.
The tradition's recurring difficulty is rigidity. Kant himself, in a notorious 1797 essay, held that one may not lie even to a murderer asking where his intended victim is hiding, and defenders have been trying to explain that away ever since. W. D. Ross responded by proposing prima facie duties which are genuine but defeasible when they conflict, and much subsequent work is an attempt to keep the constraints while allowing them thresholds.
In political thought the tradition supplies whatever grounds the claim that rights are not merely useful. It underwrites the prohibition on torture, the objection to punishing the innocent for deterrent effect, and the general suspicion of arguments that begin by aggregating welfare — which is why it functions as the standing alternative to the utilitarian framework in nearly every applied debate.