A more intelligent wheelchair

Designing a wheelchair that is able to adapt to a user’s specific disabilities, as well as his/her level of fatigue, and allowing his/her medical records to be updated at any time is the aim of the Sysiass Project (2010-2013), which brings together researchers from the Ecole centrale de Lille, the ISEN in Lille and the Universities of Kent and Essex in the UK, as well as the University Hospitals of Kent and the Institut catholique de Lille. “The English teams have expertise that complements our own.  We had been working with the University of Kent on an autonomous navigation project for mobile robots and boats, so it was quite natural to continue working together on this project and to involve researchers from the University of Essex, who are working on the design for new human-machine interfaces for electric wheelchairs,” explains Annemarie Kökösy from the ISEN in Lille, which is coordinating Sysiass.

 

We must be sure that we are meeting real needs, which is why users are involved in the project, allowing them to specify their needs and test the devices.

The main focus of the project is to be able to adjust a wheelchair’s degree of autonomy — from driving assistance to autonomous navigation — depending on the user’s capabilities, and also to ensure that the device can be fitted on standard existing wheelchairs. “We have based our approach on the fact that disabled people wish to retain control of driving their wheelchairs, as far as possible,” underlines Annemarie Kökösy . An initial driving assistance prototype has already been produced and is currently being tested by the Fondation de Garches and the CIC-IT Department of the Raymond Poincaré Hospital in Garches. This prototype slows the wheelchair down when approaching hazardous obstacles and stops it if the user has not given the command him/herself, and provides visual feedback allowing the hazard to be located. The second anticipated scenario is semi-automatic operation, with the wheelchair itself avoiding an obstacle if the user has not done anything, and automatically passing through doorways. With requirements varying significantly depending on the user, researchers have launched a European survey with the aim of identifying needs. This will allow them to incorporate all the solutions in the same device. “The results of the survey will be available in January or February, but I am already seeing initial feedback emerge for a third scenario:  the possibility of allowing the wheelchair to guide itself to the location indicated to it, by voice command, for example.”

Less progress has been made on the two other aspects of the project. One relates to the design of a secure communication module, a speciality of the team in Kent. This will allow hospital personnel and home care workers to easily share information contained in patients’ medical records and to update these records at the point of care. The other aims to invent alternative control systems to the joystick, in order to improve the autonomy of tetraplegic patients, for example. The Essex team is working specifically on a voice interface and an interface that works by detecting eye movements.

A by-product of networking by those involved in the Sysiass Project: working with medical personnel and users is revealing numerous other needs, in terms of improving the lives of disabled people, both in hospital and at home. “It offers the opportunity for engineering students to work on small projects, such as designing a voice control system for smartphones or for television remote controls.  Sysiass has launched an initiative, which may lead to other larger scale projects!”

L’automatique au service du handicap

© Inria/ Kaksonen

“I always say that control engineering develops software for hardware,” says Jean-Pierre Richard, Non-A team leader, member of the Lagis-CNRS and professor at the Ecole Centrale de Lille. “This means that control engineering is both omnipresent and invisible to users.” Therefore, you will not be surprised to discover that this discipline has also been involved in the Sysiass Project. Indeed, an automatic or semi-automatic wheelchair needs to know how to make the right movements to avoid obstacles or pass cleanly through a doorway. It also needs to be able to determine its location, detect obstacles and create a map of its environment. The first obligation falls within the scope of control theory and the second within signal processing. These are two areas in which Non-A researchers are developing theories and algorithms, the applications of which could lie equally in the management of a hydro-electric dam or in economic or robotic systems, the team’s application of choice.

 

It is very important to be able to establish a connection between our theoretical approach and its actual implementation, in an application taken as far as the prototype stage.

 

 

The Sysiass Project has already led to the publication of a significant result at this year’s ICRA (International Conference on Advanced Robotics), regarding the possibility of determining a location using a single “landmark” or reference point instead of the three previously deemed essential for calculating a position. The trick is to make use of the movement pattern (for the boat, robot or, in this case, the wheelchair), to reconstruct missing data, as well as to very rapidly reconstruct sensor signal drifts, which is one of the team’s specialities. The wheelchair is able to identify fixed points in its environment (doors, cupboards), which are all absolute landmarks but which, during the course of its movement, may be lost from sight: even then, it can continue to determine its location using only one of these points. “For us as researchers, the interesting challenge is to progress as far as making a physical version of the device.  This requires a genuine managerial link with the teams of engineers at the ISEN, which have the necessary development expertise.  This is possible for Inria as the Institute is keen to develop applications that are useful to society.”