The i-Walker: An Intelligent Pedestrian Mobility Aid - Computational Intelligence in Healthcare: Advanced Methodologies - page 102

Fig. 1. i-Walker

The agent based control system provides an excellent means to model the different
required autonomous elements in the patient's environment (from control elements in
the walker to care-giving services). Agents probe to be efficient in coordinating hetero-
geneous domain-specific elements with different levels of autonomy. Addressing the
mobility problem and keeping in mind that different users need different degrees of
help, a part of this agent based control layer has been focused on the development of a
shared control for the robotic walker that adapts to the user needs.
The i-Walker is an assistive device with four conventional wheels and two degrees
of freedom (see figure 1). Two of these wheels, the ones placed closest to the user, are
fixed wheels driven by independent motors. The other two wheels, the ones placed on
the front part, are castor-wheels. They can freely rotate around their axis and are self-
oriented. The i-Walker has two handles, that the user holds with both hands, to interact
with it. The i-Walker is a passive robot as it will only move if the user moves it.
The mechanical analysis of the Intelligent Walker is focused on the interaction be-
tween a generic user and the vehicle, in addition to how the rear wheel motors -which
are the only active control available- can modify the user's behaviour and his/her per-
ception of the followed path. For safety reasons, these motors will never result in pulling
the i-Walker by themselves.

4.1 i-Walker Control Concept

The walker has been designed to be passive, cooperative and submissive (see figure 2).

- Passive because it can only adjust the facing direction of its front wheel, i.e . it can
steer. However, it has no forward drive motors and so relies on the user for motive
force. This allows the walker to move at the user's pace and provides for the user's
feeling of control.
- Cooperative because it attempts to infer the user's path and uses this inference to
decide how to avoid any obstacles in the user's path.