XClose

UCL Division of Surgery and Interventional Science

Home
Menu

Adaptive shared control systems

Shared control systems combine automation and user control into a hybrid control system. During work experience, 17-year-old Ciara Gibbs shadowed PhD student Chi Ezeh to learn more.

Adapted wheelchair in a public place

1 August 2017

What is an adaptive shared control system?

A shared control system is one in which both automation and user control are combined into a hybrid control system. The human user of the technology will interact with automated components, usually in the form of robotics, to produce a desired outcome.

A shared control system becomes adaptive when it can adjust itself to handle varying parameters, such as the distance between the system and an obstruction. This means that that the system can effectively distribute control between the user and the automated components depending on the situation.

Consequently, the overall performance (in terms of safety, ease of use, etc) of the system can be improved.

What is wrong with commercial wheelchairs?

Current interfaces used for powering commercial wheelchairs are not sufficient for the needs of many people with disabilities.

  • Joystick: can be difficult to consistently manoeuvre in part because it is hard to know what is by the sides and at the back of the wheelchair. This interface is also inadequate for those who have upper limb paralysis.
  • "Sip/Puff" switch: users activate switches and hence wheelchair movement through puffing into or sucking out of a mouth tube. However, after prolonged use, this interface can become both tiresome and painful.

With an increasing number of people needing mobility aid, it is important that ‘smart’ devices are designed to maximise the quality of life of their users.

The experiment

Consider the movement of a wheelchair. A simple shared control system will linearly mix the user's intended direction or velocity with that proposed by a path planner algorithm (a basic navigation assistance algorithm). This does not always ensure safety, especially in cases where the user’s input is misinterpreted. The experiment will evaluate probabilistic blending, a method which aims to optimise the joint probability distribution of the user’s and path planner's decisions.

Joint probability distribution is a type of probability distribution that can define the behaviour of two random variables, X and Y, simultaneously. It can be used for both discrete and continuous random variables. In this case, the information is being used to assess the environment for obstructions and the behaviour of the user. Therefore, the probabilistic framework can then provide a suitable level of assistance which is personalised for each user, distributing the control of the wheelchair between the user and automated components uniquely to guarantee safety.

The system: Samples are taken from input data from both the user interface (in this example a headset) and from the environment from sonar sensors.

Data from these interfaces are combined using a shared control algorithm which runs in a processing board attached at the back of the wheelchair. The output of the algorithm determines the control velocities needed at that moment and hence the shared control system is adaptive.

Drawing of a headset

One quick touch of the red button with the head selects if the wheelchair will move forward or backwards. A longer touch of the red button will pursue forward or backward motion. The yellow and blue buttons are for moving either right or left.

The digital interface uses sonar sensors, which are low cost. Sonar sensors can measure the distance between the wheelchair and an obstruction by using sound waves. These were then modified to add infrared sensors. The emission and detection of infrared waves enables the detection of obstructions. The combination of both sensors aims to ensure collisions are avoided.

Sonar and infrared sensors mounted on the wheelchair

Demonstration

I was able to try out the wheelchair. When intentionally moving into an obstruction, I found that the interface successfully prevented me from colliding and getting injured. The sensors attached to the wheelchair stopped the wheelchair within roughly 10 inches of a cardboard obstruction.

If I persisted in trying to move forward, the wheelchair would only move in short increments of roughly one inch. If used in the outside world, this technology could reduce wheelchair associated injuries as even if a collision were to occur, the impact force would be minimal.

The project has seen some promising results. The work is a great step towards perhaps bringing affordable smart wheelchairs on the market soon.

Instrumented wheelchair for shared control

Ciara Gibbs

About Ciara

I am an A-Level Student from London. In 2017, I spent time at Aspire Create doing work experience. It was fascinating to see the work the researchers are doing to improve the lives of people. I saw projects involving smart wheelchair devices, exoskeletons, neuromodulation, implants, prosthesis and many more. It is truly inspiring to witness the efforts of ordinary people who wish to make a difference.