ON-LINE SENSOR FAULT DETECTION SYSTEM FOR A SMART WHEELCHAIR Gianluca Ippoliti ∗ Sauro Longhi ∗ Andrea Monteri` u ∗ ∗ Dipartimento di Ingegneria Informatica, Gestionale e dell’Automazione Universit`a Politecnica delle Marche Via Brecce Bianche, 60131 Ancona, Italy Abstract: This paper presents a set of results on the design and on the experimental validations of a Sensor Fault Detection System (SFDS) for a smart wheelchair to prevent simples navigation sensor faults and hence to reduce the risk of safety hazards. The design has been performed by the Structural Analysis technique which uses graph theory to determine which redundancy exists in the sensory system and thus shows the possibility to detect and handle the navigation sensor faults. The considered navigation sensor equipment includes odometric, gyroscopic and laser range measures. The experimental tests have been developed by a smart wheelchair equipped with a navigation system. Copyright c  2006 IFAC Keywords: Fault diagnosis, fault detection, fault isolation, wheelchair, analytical redundancy relations, structural analysis. 1. INTRODUCTION Assistive Technology is basically the use of tech- nology to assist an individual with physical dis- abilities to perform functions that may otherwise be difficult or impossible. In this context, robotics plays a major role thanks to the large number of aids it provides for the persons with disabil- ities (Dario et al., 1996; Robinson, 1993). Many activities in this sector have been developed in the European Union within the TIDE initiative (TIDE, 1993). In this framework, the develop- ment of intelligent navigation systems based on sensors that are easily ported between various electric vehicles has been performed in the TIDE project SENARIO (Dallaway et al., 1995; Katevas et al., 1997). Another TIDE project, MOVAID (Dario, 2000), mainly focused on the develop- ment of a modular mobile robot interacting with the user by means of a friendly interface for the accomplishment of domestic tasks. Other significant research activities are those related to smart powered wheelchairs. In the developed smart wheelchairs, the autonomy and safety as- pects are guaranteed by a set of sonar sensors. Different sensor devices, such as vision systems, are also used (Gomi and Griffith, 1998; Wheeles- ley, 1997; Levine et al., 1999). This paper presents the design of a Sensor Fault Detection System (SFDS), its integration with the navigation system for a commercial powered wheelchair and the related experimental valida- tions. The aim is to prevent that simples nav- igation sensor faults develop into serious fail- ure and hence increase the wheelchair availabil- ity and reduce the risk of safety hazards. This is accomplished applying the Structural Analysis technique which uses graph theory to determine which redundancy exist in the system and thus shows the possibility to detect and handle the navigation sensor faults as stated in (Declerck and Staroswiecki, 1992; Blanke et al., 2003b). The Sensor Fault Detection System provides the commercial wheelchair with a set of functions which increase the navigation reliability of the wheelchair, and thus the autonomy of elderly and people with motor disabilities. The developed ex- perimental tests provides a significant validation of the detection system.