Human Intention Recognition Based on Markov Decision Processes

Ubrzan rast e-trgovanja povećava potrebu za većim skladištima i njihovom automatizacijom. Korištenje robota za asistenciju ljudskim radnicima postaje prioritet. Robot mora prepoznati čovjekove namjere ako želi asistirati efikasno i sigurno. Teorija uma je ljudski intuitivni koncept procjena mentalnih stanja drugih agenata, poput uvjerenja i želja, i kako ona utječu na ponašanje. U ovom smo radu prezentirali algoritam za prepoznavanje čovjekove namjere baziran na teoriji uma koristeći Markovljeve procese odlučivanja i skriven Markovljev model. Pokazali smo kako predložen algoritam precizno predviđa ljudske želje.The rapid growth of e-commerce increases the need for larger warehouses and their automation. Using robots as assistants to human workers becomes priority. In order to operate efficiently and safely, robot assistants must recognize human intentions. Theory of mind (ToM) is our intuitive conception of other agents' mental state, i.e. beliefs and desires, and how they cause behavior. In this paper we present a ToM-based human intention recognition algorithm. We have placed the warehouse worker in a 2D simulation environment with three potential goals.We observe agent's actions and validate them with respect to the goal locations using Markov decision process framework. Those observations are then processed by proposed hidden Markov model framework which yields agent's desires estimations. We show that the proposed framework predicts human warehouse worker's desires accurately and comment upon the simulation results