Research On Asynchronous Brain-acuated Control Of Mobile Platforms

Brain computer interface(BCI)is a new type of human-computer interface(HMI)technology which provides a direct communication and control pathway between the brain and the external world,bypassing the participation of peripheral nerves and muscles.A BCI can therefore serve as a replacement or restoration of natural central nervous system(CNS)output,and thereby help people with severe neuromuscular dysfunction regain their ability to communicate and control.With further BCI research,its application also benefits healthy individuals.How to enhance,supplement,or improve natural CNS output for able-bodied people has also been widespread concerned.BCI research is now an explosively growing field involving dedication of hundreds of research groups throughout the world.The research filed of BCI technology has evolved from basic communication functions to complex human-machine coordinated control system.Its application has also ranged from designing spellers to realizing more complicated and practical control system,such as robots,wheelchairs,prostheses.The past decades have witnessed many impressive achievements made by BCI technology.Various prototype demonstration systems have emerged and well illustrated the promising prospects of using BCIs to control external devices.But there are still several bottlenecks,such as unsatisfactory signal acquisition hardware,limited degrees of freedom(DOF),and the relatively poor robustness of the system,which will all hinder the practicality,especially its application,--of BCI technology in the future.This paper aims to build brain-controlled mobile platform systems,including a BCI-based wheelchair and a car.Our efforts to enhance the practicability of BCI system are mainly made in two aspects: i)woking on asynchronous brain-acuated control of mobile platforms,to achieve more practical,user-friendly and natural human-machine interactions;ii)increasing the DOF of brain signals,which enables the BCI system to control complex devices,like wheelchairs and cars,with more commands available.The outline of researchs and results are as follows:The P300-BCI makes it possible to detect user intent by selecting the target in multiple options displayed on the screen.By mapping different target options to specific control commands,the P300-BCI can realise basic functions like communication and device control.The P300-based speller,enabling individuals with severely kinetic disorders to communicate with the outside world,is one of the typical BCI applications for multi-target operation,which can be used to evaluate the performance of a multi-target operating system by considering the accuracy,the information transmission rate,etc.Currently,there is still a general lack of BCI system designed for inputting Chinese sinograms,which include more than 11,000 commonly used characters in the library.Considering the large scale of this problem,we presented a P300 visual Chinese char-acter spelling system with a Hanyu Pinyin-based method.Ten healthy subjects participated in the study and achieved an average offline accuracy of 92.6% with a mean information transfer rate(ITR)of 39.24 bits/min and an average online input speed of 1.4 sinograms per minute(with an accuracy rate of 100%).The preliminary results presented here indicated that the online input of Chinese text using a Pinyin-based visual speller is feasible.In this paper,the efficiency and practibility of P300-based multi-targets selecting BCI was also evaluated to make sure this kind of BCI is suitable for further implementation in hybrid BCI system.Motor imagery(MI)-based BCI is an important component of BCI technology.The small amount of independent commands provided by MI-BCI is far from enough to control complex devices like motor cars.To tackle this problem,we proposed a multi-class BCI paradigm to control a car.This paradigm,extending available commands while without occupying the user's visual and auditory pathways,uses two distinct MI tasks,namely imaginary left-and right-hand movements.These commands were used to generate a multi-task car control commands consisting of starting the engine,moving forward,turning left,turning right,moving backward,and stopping the engine.Five healthy subjects participated in the online car control experiment,and all succeeded in following a previously defined route.This experiment pictured a promising blueprint for individuals with locked-in disorders to gain more mobility in the future,,as well as provided a supplementary car-driving strategy for able-bodied people.How to improve the operability of BCI system and how to design friendly human machine interface are two important problems to be solved when it comes to achieving truly practical and effective BCIs.In this paper,we proposed a self-paced hybrid BCI paradigm by merging a motor imagery(MI)-based brain switch into a P300-based BCI approach.This work not only tested the system performance,but also proved its availablity of asynchronous operation.The practicability and effectiveness of the above-mentioned paradigm was validated by eleven participants,and all of them achieved a satisfactory performance.An average online input speed of 7.8 characters per minute(with an accuracy rate of 100%)was achieved.The paradigm capacitated users to voluntatily decide on whether to give commands or not,how long to take control and when to stop,thus making it more practical for daily using scenarios.EEG-based brain-controlled wheelchair is one promising and typical application of BCI.In this paper,we proposed a hybrid BCI paradigm that combines MI with P300 potential for the asynchronous operation of a brain-controlled wheelchair.This paradigm is completely user-centric.By sequentially performing MI tasks or staring at P300 flashing stimulis,the user had access to twelve functions to control the wheelchair: start,move forward/backward,move left/right,move left45/right45,accelerate/decelerate,turn left/right and stop.The proposed approach achieved good performance,as all subjects can control the wheelchair to reach the destination smoothly without neither ex-ternal intervene nor additional assistance.In current developing stage,MI-based BCI is still not competent to perform complex control tasks.One of the main constraints is limited available degrees of freedoms(usually 2 to 4),which cannot meet the DOF requirements for complex systems.In this paper,we presented the preliminary experimental results of single-trial classification of four types of MI signals differing in task complexity(i.e.,simple versus complex tasks)using a novel empirical mode decomposition & common spatial pattern((EMD-CSP)feature extraction approach.The experimental results revealed that an average offline accuracy of 71.56% was achieved over 8 participants,implying that separable differences exist among the four types of MI tasks with varying task complexity,which might encourage further efforts for the realization of a multiclass MI-based BCI paradigm.However,the effectiveness of the method,as well as online performance also needs further validation.