Descrizione del progetto
Un’iniziativa di sensibilizzazione alla ricerca e all’innovazione in Serbia
L’esplorazione robotica dello spazio è essenziale, in quanto consente di studiare ambienti troppo pericolosi e lontani per gli esploratori umani; sfortunatamente, tuttavia, in tali contesti possono originarsi gravi e irreparabili malfunzionamenti dell’hardware, che spesso portano alla disattivazione dei robot. Riconoscendo questa inefficienza, il progetto RECOVER.ME finanziato dal CER, si propone di sviluppare una soluzione innovativa per la gestione dei guasti basata sulla metacognizione, in grado di emulare la comprensione umana dei guasti. Questa soluzione consentirebbe ai robot di riprendersi da malfunzionamenti hardware gravi ma correggibili senza alcuna assistenza esterna, consentendo di incrementare l’efficienza dei futuri progetti di esplorazione e infrastruttura spaziale.
Obiettivo
The aim of the RECOVER.ME project is to achieve human ingenuity in dealing with hardware faults in robotic space exploration. The hypothesis of the project is, that as robots acquire human-like metacognitive
awareness and metacognitive regulatory abilities, they will be enabled to recover from severe but rectifiable hardware malfunction all by themselves. This is of particular importance to planetary exploration, as
a hardware fault need not be the end of a mission. However, as of today, once a hardware malfunction occurs, the remote robot is typically taken out of operation and troubleshooting is done manually. In the
future, especially, when more complex robots are deployed to construct planetary infrastructure for crewed exploration, this can no longer be tolerated. Considering that a hardware fault may occur at any time, such
a situation can become safety-critical for the robot, the established infrastructure, and for astronauts in the vicinity of the robot.
To overcome this issue, RECOVER.ME proposes a novel approach for metacognition-enabled failure handling. Instead of relying on hard-coded recovery strategies by specifying how a robot has to react to a certain sub-system fault, the project aims to bootstrap failure handling as a property of the cognitive architecture of the robot itself. Metacognitive awareness is created through a novel knowledge representation that describes how hardware faults may impact robot capabilities. Metacognitive planning will yield contingency configurations employing abstract, affordance-based first order-logic planning for self-programming. To empower robots to monitor their own programming and evaluate the best strategy to react to arbitrary failure cases, generic limitation models will translate sub-symbolic fault information into semantically interpretable knowledge for metacognitive monitoring and metacognitive evaluation. This will provide robots with competent strategies to deal with faults in a similar way to humans.
Parole chiave
Programma(i)
- HORIZON.1.1 - European Research Council (ERC) Main Programme
Argomento(i)
Invito a presentare proposte
(si apre in una nuova finestra) ERC-2023-STG
Vedi altri progetti per questo bandoMeccanismo di finanziamento
HORIZON-ERC - HORIZON ERC GrantsIstituzione ospitante
51147 Koln
Germania