New approach for brain-controlled wheelchairs could achieve 99% driving accuracy

February 23, 2021
A brain-computer interface might control new wheelchairs. (Pixabay/Kevin Phillips)

A brain-computer interface might control new wheelchairs. (Pixabay/Kevin Phillips)

A Portuguese study has shown for the first time that it is possible to achieve a highly reliable and nearly 100% accurate brain–computer interface system for use in brain-controlled wheelchairs, without imposing a high mental workload on the user.

A brain–computer interface, or BCI, is a direct communication pathway between a wired brain and an external device that makes it possible to send commands through brain signals without requiring muscle activity. These interfaces have many clinical and non-clinical applications, including motorized wheelchairs, Gabriel Pires, a researcher at the Institute for Systems and Robotics of the University of Coimbra in Portugal and a co-author of the paper, told The Academic Times.

In the paper, published Jan. 27 in the IEEE Transactions on Human-Machine Systems, a team of researchers in Portugal tested different approaches for controlling motorized wheelchairs in people with physical disabilities, including a new adjusted-time commands approach, which adjusts the brain–computer interface speed to the user’s performance over time. Those with severe motor impairments can often struggle to use traditional interfaces in powered wheelchairs that require muscle movement, and brain–computer interfaces may offer an alternative solution.

Typically, a brain-controlled wheelchair functions using a brain–computer interface in the form of electroencephalogram electrodes worn on the user’s head, which detect neural impulses and the user’s thought process to control the wheelchair’s movements.

However, using a brain–computer interface to control a robotic wheelchair is a challenging task, the authors said, because brain–computer interfaces have low transfer rates, meaning the speed at which information transfers from a person’s brain to the device, and limited accuracy. Operating the system requires continuous attention and focus, which can lead to high mental and physical workloads for the user.

“The use of BCI in real environments is not yet a reality, and BCI is very much confined to laboratory experiments and controlled environments. One of the main challenges that BCI faces is the achievement of high reliability and accuracy,” Pires said.

Brain-controlled wheelchairs must be highly reliable for safety reasons. Pires explained that if a car's brakes only work properly 90% of the time, and sometimes it takes one second to brake and sometimes it takes 10 seconds, for instance, that car would be deemed unreliable and dangerous to drive. The same is true for brain-controlled wheelchairs, which additionally require a great deal of focus to operate.

But the reliability of brain-controlled wheelchairs increases when they integrate an assistive navigation system that can perceive the wheelchair’s surroundings and perform corrective, autonomous movements. This combination of user intent and navigation information is called a “collaborative control” condition, Pires said, and it is able to disregard wrong commands from the user and replace them with the right ones.

For the current study, the research team recruited seven non-disabled participants and six participants with severe motor disabilities. The researchers tested a combination of approaches for more reliably controlling robotic wheelchairs, including collaborative control, and how natural the machine interactions were for users.

Participants were instructed to navigate three pathways while operating a robotic wheelchair with their brain in office-like environments. For example, one task included three narrow doorways, two small obstacles and two large obstacles. The minimum number of directional decisions the participants had to make in order to reach the final destination was five.

A digital screen attached to the robotic wheelchair displayed seven steering command options: forward, back, left 90 degrees, right 90 degrees, stop, WC (meaning toilet) and help. Prior to navigating the paths, participants completed a calibration session in the wheelchair to familiarize themselves with the system. Each task tested the participants with different conditions.

The most successful approach for the participants was a combination of self-paced control, adjusted-time commands and collaborative control. In self-paced control, directional commands are able to be sent naturally at the user’s own pace; in adjusted-time commands, the detection time for commands given by the brain can be adjusted to allow for potential distractions, decreasing the users’ performance fluctuations.

Non-disabled and motor-impaired participants used the self-paced brain–computer interface with a mean accuracy of 95.8% and 93.7%, respectively. The collaborative control condition increased the overall system accuracy to above 99% for both groups. Even when using the non-self-paced approach, collaborative control increased the overall accuracy to 94.6%. 

“This was an important achievement, although we are aware that these experiments took place in a relatively controlled environment, much less complex than domestic environments and what daily tasks require,” Pires said.

The results achieved by the motor-disabled participants suggest that this particular brain–computer interface model may be an effective solution for wheelchair control. The technology is not ready for commercial use yet, the authors advised, but the findings are an important step in that direction. 

The researchers are already working on the integration of vision sensors in the system to recognize obstacles like doors, tables and chairs. And Pires suggested that future studies on brain-controlled wheelchairs should involve end-users in the research and development process and run more extensive experiments with a wider group of participants.

“Our work presents the most complex navigation scenario including both healthy and severely motor-disabled participants,” the authors said in the paper. “To the best of our knowledge, our proposal is the only one achieving an overall accuracy greater than 99%, which validates the proposed BCI and navigation approaches.”

The study, “A Self-Paced BCI With a Collaborative Controller for Highly Reliable Wheelchair Driving: Experimental Tests With Physically Disabled Individuals,” was published Jan. 27 in the IEEE Transactions on Human-Machine Systems journal. The co-authors of the paper were Gabriel Pires, a senior researcher at the Institute of Systems and Robotics (ISR) of the University of Coimbra and a professor at the Polytechnic Institute of Tomar (IPT); Aniana Cruz, a Ph.D. student at ISR; Ana Lopes, a researcher at ISR and professor at IPT; Carlos Carona, a researcher at the University of Coimbra; and Urbano Nunes, a senior researcher at ISR and professor at the University of Coimbra.

We use cookies to improve your experience on our site and to show you relevant advertising.