dcyphr | Human machine interface: robotizing the instinctive living


The continuous improvement of Brain Computer Interfaces (BCI) displays incredible potential. This study reviews modern techniques, their respective sensors, and their commercial viability. This review of the BCI technologies and their sensors could be of use in demonstrating the importance of brain machine interfaces. 


A Brain Computer Interface (BCI) outfit 

BCI research aims to augment or repair human sensory motor or cognitive abilities with the development of advanced technologies. BCI systems must first sense the brain signals via scalp, brain surface electrodes, or from direct neural activity. Then these signals are sent to several computers for processing and to remove background noise. These signals are then sent to a robotic arm, wheelchair, or any variety of robot that helps the user traverse and navigate the environment around them. 

Paper Organization 

This paper describes either non-invasive or invasive sensors. Invasive sensors are directly embedded into the brain by surgery. Non-invasive sensors are applied to the scalp or on top of hair without the assistance of surgery. 

BCI technologies 

EEG- Electroencephalography is a non-invasive technique that can detect brain activity by the application of electrodes to the scalp. It is the most commonly used system. EEG is fast and cheaper than other methods available. The largest issue is that in order to get information required the Region of Interest (ROI) on the brains’ surface is required. Otherwise the neuronal activity recorded will not be of interest. 

ECoG-Electrocorticography is an invasive technique that requires surgery to insert sensors into the brain. It has been used on animals and humans. Its largest issue is that ECoG requires expensive specialty equipment. The signal acquired from the sensors is similar to that of EEG. The largest difference is that the sensors are implanted into the brain. 

ECG-Electrocardiography is a non-invasive technique where the electrical activity of the heart is recorded by external monitors placed on the chest of the human/animal. The largest issue is that this technique can only be applied to monitor the heart activity. 

EMG-Electromyography is a noninvasive technique where the sensors are placed on the muscles of the body. It is used to measure electrical activity of skeletal muscles. It is quicker than EEG and can be used to induce specific information about neuronal activity. For example if small muscle contractions were observed than they must be caused by lower amplitude electrical signals from the brain. 

MEG-Magneto Encephalography is a noninvasive technique where the brain signals are recorded using magnetic fields generated by the current present inside the brain. The main downside is that this requires equipment that can detect the small signals from the brain while simultaneously protecting from magnetic fields produced from the environment. It requires a large room that is insulated from magnetic fields.

fMRI-Functional Magnetic Resonance Imaging is a non-invasive technique where the brain signals are understood by measuring the blood flow inside the brain. This operates under the assumption that brain activity and blood flow are coupled. It is able to create high resolution images of the brain. 

NIRS- Near Infrared Spectroscopy is a non-invasive technique. It uses electromagnetic spectrum (800nm to 2500nm) to monitor electrical activity of blood and sugar levels. It is incredibly expensive and is under current development. 

Review of Sensors for BCI

There are three types of implantable microelectrode arrays. The first is known as the Utah array which is silicon based and signals can only be collected from the tip of each electrode. There are 100 needles and it is stiff. Thus information carried from it is very limited. The second type is known as the Michigan array. The Michigan array is also silicon based,  but a major advantage is that it is not fixed to any specific organization. It also offers greater resolution compared to the Utah array. The third type is known as flexible arrays. All of these sensors can be utilized during ECoG. 

    Local Field Potential (LFP) will monitor the brain through non-invasive methods. It can utilize capacitance or non-capacitance based electrodes. Wet Electrode Arrays (WEA) use a special gel to get better brain signals but require the user to wash or remove their hair. They are used in EEG and MEG techniques. The Hybrid dry Electrode Sensor Array (HESA) achieves the same sensitivity and resolution as the WEA, but does not require gel and can work through hair. Superconducting Quantum Interference Device (SQUID) Magnetometers are used in MEG. It requires a room insulated from magnetic fields and another room dedicated to computation. It is not used often due to its high cost. 


Remote-brained humanoid

This is a strategy developed back in 1998 when computers were less efficient and much heavier. This strategy removed the processing from the robot and placed it on an external computer. This way the robot had to move less weight and could move with more digress of freedom.

Electrode for sensory motor

A Ceramic Based Multi-Site (CBMS) electrode was developed at Drexel University in 2001. This electrode could produce limb movement by utilizing spinal cord neurons. It was designed for prosthetic devices to restore sensory feedback. 

Continuous shared control system for BMI

This gives the control of the robotic system to two different networks. One network is deemed the “reflex” and is operated by simple control networks informed by specific sensors. The second network that worked in conjunction with the reflex was one controlled via the brain. This offered significantly better control of the robotic arm. 

Motor Imagery based BMI

Utilizes an EEG to control a robotic arm. Different states of brain excitation are established and then used as a comparison for controlling the robotic arm. This technique was used to permit a paraplegic to control his motorized wheelchair again.