Periodic Reporting for period 1 - BrainControl (Stable Brain-Machine control via a learnable standalone interface)
Okres sprawozdawczy: 2016-09-01 do 2018-02-28
Brain Machine Interfaces (BMI) have brought unparalleled promise of autonomy for disabled people through the direct control of devices using brain activity, with large impact in the areas of neuro-rehabilitation and neuro-prosthesis. Non-invasive BMIs, in particular, can also be helpful for both healthy and aged users to control everyday devices and perform simple tasks.
We developed a prototype for a novel, stand-alone, noise-resistant, multi-platform BMI, based on a novel learning-based approach, in which each subject learns to use a fixed decoder to control an actuator. This approach takes advantage of the brain’s unique ability to learn transformations between brain activity and actions, similar to what happens when we learn motor skills, for example.
We have implemented an adaptive decoder into a programmable hardware-based solution, which provides complete portability and a cleaner and simpler interface. We have successfully transitioned from a setup requiring a processing computer to analyze the neuronal data, to a system that can automatically implement our adaptive algorithm, while still allowing for minor adjustments to be introduced as required.
We have also implemented a Virtual Reality (VR) game environment to promote a more immersive and interactive training protocol, allowing the user to more rapidly and intuitively gain proficient control of the BMI device.
Finally we have partnered with Darwing to develop a study to evaluate the current BMI market scenario and market sectors possibilities from health to gaming, and from neuromarketing to control. This has helped in identifying possible commercial applications, define our geographical IP protection and outline a business plan and commercialization strategy.
We developed a prototype for a novel, stand-alone, noise-resistant, multi-platform BMI, based on a novel learning-based approach, in which each subject learns to use a fixed decoder to control an actuator. This approach takes advantage of the brain’s unique ability to learn transformations between brain activity and actions, similar to what happens when we learn motor skills, for example.
We have implemented an adaptive decoder into a programmable hardware-based solution, which provides complete portability and a cleaner and simpler interface. We have successfully transitioned from a setup requiring a processing computer to analyze the neuronal data, to a system that can automatically implement our adaptive algorithm, while still allowing for minor adjustments to be introduced as required.
We have also implemented a Virtual Reality (VR) game environment to promote a more immersive and interactive training protocol, allowing the user to more rapidly and intuitively gain proficient control of the BMI device.
Finally we have partnered with Darwing to develop a study to evaluate the current BMI market scenario and market sectors possibilities from health to gaming, and from neuromarketing to control. This has helped in identifying possible commercial applications, define our geographical IP protection and outline a business plan and commercialization strategy.