Some studies show that 40% of patients who use a power wheelchair think it is difficult to drive, and between 5% and 9% require the assistance of another person [2]. Sometimes it is possible to improve the autonomy with smart wheelchairs that are equipped with sensors and exploit control algorithms from the field of mobile robotics [3]. This is not possible in other cases, so an attendant to drive the chair is required. In these cases, the chairs usually have so-called ��joysticks�� which are incorporated in the rear handle of the chair [4].However, as already mentioned, these human-machine interfaces are not always the most appropriate tool. Some authors propose the use of a tactile screen instead [5,6]. The user touches the screen, similar to that of common smart devices nowadays, to lead the chair or indicate the destination.

However, this solution often suffers from the disadvantage, described above, of the lack of easy handling and thus the incorrect interaction between man and machine, which is detrimental to the usability of the system [7].Therefore, there is a need for an intuitive and highly usable interface for power wheelchairs. This interface can also be used to drive electric trolleys, for instance shopping trolleys [8], which will also increase and extend the autonomy of the elderly or impaired. They will even be helpful for healthy people that have to drive trolleys with heavy loads [9]. In [8,9] it is proposed that the user takes a handle or grip with his or her hands and drives the vehicle as if he or she were pushing it.

However, they do not really use GSK-3 a direct interface between man and machine, moreover the force actually exerted on the handle is transferred to other points where force and/or torque sensors are placed. Such action can lead to problems arising from the deformation of the elastic bands or the means used to transmit the user commands. Furthermore, the deformation of these areas can have a complex behavior over time as a result of a complex dynamics of the set, and the readings of the sensors placed in the bands or elastic joints should be registered, which makes the control (from the readings of the sensors) complex. Another interesting work devoted to exploring the role of a tactile sensor as a human-machine interface is presented in [10], where manipulative forces are predicted from the information provided by the tactile images of the forearm.

The proposed device is based on a tactile sensor [11] to achieve a direct interface with the attendant. The tactile sensor provides information not only about the location of the contact points, but also about the contact force at these points, so its output is a force map or tactile image. The processing of this map gives the information required to drive the chair or trolley in an intuitive way.