You are here: News Robot wheelchair with EI instead of GPS

Robot wheelchair with EI instead of GPS

A likely consequence of the senescence of our society is that more and more people will require a wheelchair. This subject fascinates engineer Alexander Hüntemann (second from right on the photo), who is designing an understanding wheelchair. By taking the steering skills and behaviour of the user into account, the robot can increasingly fine-tune the wheelchair’s route.

A likely consequence of the senescence of our society is that more and more people will require a wheelchair. This subject fascinates engineer Alexander Hüntemann (second from right on the photo), who is designing an understanding wheelchair. By taking the steering skills and behaviour of the user into account, the robot can increasingly fine-tune the wheelchair’s route.
Robot wheelchair with EI instead of GPS

© Rob Stevens - KU Leuven

Hüntemann took his engineering degree in Madrid, came to Leuven with a Marie Curie scholarship and worked at LMS International, a spin-off of KU Leuven that develops simulation and test software in the mechanics industry. From LMS, he moved to the Department of Mechanical Engineering at KU Leuven. Supervised by Professors Hendrik Van Brussel and Marnix Nuttin, he defended his doctorate on Probalistic Human-Robot Navigation on 16 March.
 
Robotics is a multidisciplinary topic, Hüntemann explains: “It concerns sensors, maths, artificial intelligence and users. The technology for mobile robotics is ripe, but we need to learn more about users and their interaction with robots. Just as there are already autonomously functioning vacuum cleaners and lawn mowers, we are able to make perfectly autonomous wheelchairs that use a GPS system to find their way. But people also like to maintain a sense of control. It is important, moreover, that they stay fit mentally and physically. That is why we are endeavouring to ensure shared control: shared autonomy between the user and wheelchair.”
The understanding robot
The point is to find a balance between existing alternatives: a manually operated or push-wheelchair; an engine-driven electric wheelchair that is steered with a joystick but is difficult to control; or a robot that steers completely independently without the user having to do anything. It is perfectly possible to design a robot that maps its environment – both indoors and outdoors – and drives through it without crashing. The challenge now is to create a robot that takes the wheelchair user into account. Every wheelchair user is different: ranging from somebody who makes uncontrolled movements to an old person who has little power in his/her hands as a result of chronic joint rheumatism. Each individual case may also change, such as when illnesses worsen, for example.
The robot examines its environment using sensors that are comparable to parking sensors on a car. The wheelchair user can steer the robot in various ways. In another project, led by Professor Nuttin, engineers are already developing a wheelchair that is steered via brain signals. The user was given a sensor cap – a kind of bathing cap with electrodes. Sensor caps are not very practical, however, and they cause quite a lot of static interference. Hüntemann’s wheelchair has a joystick because it ensures more control: “In the long term, this provides opportunities to apply haptic feedback: technology that communicates with the user through position and power. The robot can also move the joystick and let the user know which direction it wants to go in. This allows the user to correct what the robot will do before it goes in the wrong direction.”
Local routes
Hüntemann’s robot chair first learns the user’s steering skills during a training period, so that it is able to evaluate when, for example, the left hand is used less. Eventually, when the user is ready to drive the wheelchair, Hüntemann stops the GPS system in which the user puts in a destination and the robot calculates a route. Instead, Hüntemann applies a local approach: the robot only maps its immediate environment and all the smaller possible routes. The great advantage of this system is that there are many more alternative routes between two points. Every time the user moves the joystick, the robot recalculates the most probable route intended by the user. By taking all the possible safe routes and past steering behaviour into account, the robot can increasingly fine-tunes its routes.
 
Hüntemann collaborated with the Vilans knowledge centre in the Netherlands to test his prototype with various wheelchair users in a domestic test space. “The reactions were very positive. The participants asked when the wheelchair would be on the market, which depends on the manufacturers. The sensors we use are very expensive – though they are getting cheaper. Usually, wheelchairs are refunded by health insurance companies.”
 
After his doctorate, Hüntemann will continue his research into wheelchairs in the European project RADHAR (Robotic ADaptation to Humans Adapting to Robots) directed by Professor Joris De Schutter. The objective of the project is to conduct detailed investigations into the interaction between robots, people and wheelchairs in the long term.