Content uploaded by Camilo Andrés Cáceres Flórez
Author content
All content in this area was uploaded by Camilo Andrés Cáceres Flórez on Sep 13, 2016
Content may be subject to copyright.
First Submitted version without corrections
For the final vesion check
http://www.praiseworthyprize.org/jsm/index.php?journal=ireme&page=article&op=view&path%5B%5D=19047
International Review of Mechanical Engineering (I.RE.M.E.), Vol. xx, n. x
Manuscript received January 2007, revised January 2007 Copyright © 2007 Praise Worthy Prize S.r.l. - All rights reserved
Approach to Assistive Robotics based on an EEG sensor and a 6-DoF
robotic arm
Camilo Andrés Cáceres1, João Maurício Rosário2, Dario Amaya 3
Abstract – This work presents an approach to Assistive Robotics based in a Brain Computer
Interface (BCI) on the Steady State Visual Evoked Potentials (SSVEP), using a 6-DOF Robotic Arm,
an Emotiv EPOC sensor and a Kinect camera. The proposed architecture has the objective of
integrating different systems, with the goal of creating a robotic assistant for the disabled or
movement restricted user, improving the life quality of those people. The robot assistance
assignment is related with actions like pick and place, or feed and hold food for the user, etc. The
proposed methods are based in image and signal processing, BCI decision making, localization of
a target point, trajectory planning for the robot arm, and execution of the assistive task. The most
remarkable results validate the proposed approach as a good integration method of different
technologies for assistive robotics tasks. Copyright © 2016 Praise Worthy Prize S.r.l. - All rights
reserved.
Keywords: Assistive Robotics, EEG, Brain Computer Interface (BCI), SSVEP, Mechatronics
I. Introduction
Nowadays, with the medicine and public health
improvements, life expectancy is longer for people [1], [2]
and because of that, the increasing need of the assistive
technologies has grown up through the last decades. By
another hand, the elderly are not the only population that
requires this kind of assistive technology, others include
motor disabled patients, as locked-in or spinal cord injured
patients who requires this assistive technology for
improvement of their life quality.
Solutions that assisted disabled or non-disabled people
with a BCI, the main integration includes the use of a hand
orthosis [3], [4] displacement of a wheelchair [5]–[11],
rehabilitation devices [12]–[14] or the use of a robotic
device [15]–[17]. A common factor in the cited works is
the interest of improving the life quality of impaired
individuals using the different fields of engineering with a
clear objective.
In [18] they presented the results of a work using EEG
signals, and visual stimulus for an user login and
password. Based on the generated stimulus, it was
possible for the user authentication by the wavelet analysis
in sub-bands. Finally, it was possible to obtain the
matchless characteristic of a user through the application
of the discrete wavelet transform (DTW). Similarly, in
[19] a work was developed based on the movement
intention using EEG and superficial electromyography
(sEMG) signals, with the purpose of developing a
rehabilitation method using hybrid robotics. In the
developing of the work, it was possible to restore the
control of some neuromotor functions in patients with
neural injuries and a Smart Walker equipment. This
experiment achieved the detection of some brain signals,
which are related with the human body motion, as a
contribution for future works.
Following the methodology of the assistive robotics
technology commanded by a BCI, the present work
integrates a robotic arm with a BCI system in a prototype
of an assistive robot.
The main objective is to assist a disabled user with
basic daily life tasks. In this case a pick and place task was
chosen, what could be replaced in the future for a task of
feed and hold food for the user or another related assistive
task. The proposed task was exemplified with some color
cubes, located in an objective area. The task is performed
by a six degrees of freedom robotic arm, which follows
the taken decisions by the user using a BCI system, based
on SSVEP. The objective environment is supervised by a
Kinect camera. Each step of the procedure is integrated
with the objective of obtaining an approach to assistive
robotics.
The paper is organized as follows: First, the materials
and methods used for the development of the work, in
order to propose a base for the integration of each
procedure and method. Finally, the results of the work, the
full-integrated system, and the most important results of
each analyzed subsystem and the conclusions of the
proposed method.
II. Materials and Methods
The proposed robotic assistance architecture is based
on different subsystems, which integrated can help the
disabled user, with motion restrictions, with pick and
place tasks, or even grasp and serve, all by using a robot
arm. The system is shown in Fig 1, where it is possible to
appreciate the different subsystems.
Cáceres, C.A., Rosário, J.M., Amaya, D.
Copyright © 2016 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol. xx, n.
x
Fig 1. General view of the proposed architecture
The first subsystem is the Kinect Camera, which is used
for detecting the objects in the interest area and localize
them. The next element is the EEG sensor, Emotiv EPOC,
the one thru Steady State Visual Evoked Potentials
(SSVEP) allows to choose the interest object. The last
component is a robotic arm with 6 degrees of freedom
(DOF), and has the task of interacting with the
environment and assisting the user.
The order of the logic algorithms and processes follow
Fig 2, where all the processes start with the detection of
some objects using the Kinect Camera. Afterwards, with
the object selection by the user thru the BCI system, the
object location is calculated based on the Kinect Camera.
Subsequently, with the target points and using a trajectory
generator for a nonspecific and simulated robotic arm of
6-DoF, the robot can complete its task, assisting the
human. Each process has a deeper description on the next
sections, grouped by sensors.
Fig 2. Proposed Architecture Flowchart
II.1. Kinect Camera
The use of the Kinect camera sensor has two objectives.
The first one is based on use of the RGB camera, for
detecting the available objects in the interest area, and the
second is with the RGBD image, for localizing the object
position in the coordinates X, Y and Z, in respect to the
known Kinect position.
The images are taken and processed with Microsoft C#
Express 2015, the library for image processing
AForge.NET for C# and Matlab. The object detection is
based on the general analysis proposed by [20] and
supported by [21], where the image is processed using the
HSL color space and morphological transformations.
The objective detection zone is a black area with color
cubes, which will simulate the interest objects, in reality it
could be food or a bottle of water, a useful object to the
user.
The used method follows the steps and specifications
shown in the Table 1. The values were calibrated manually
for the given conditions of the environment. A sample
image transformation between each step, starting with a
RGB image of the object area and finishing with the
centroid positioning and labeling.
TABLE I
IMAGE PROCESSING METHOD AND SPECIFICATIONS
The distance calculated to the interest object is based
on the [22], [23] with the mathematical equations (1),
where the position (X, Y, Z) measured from the Kinect
sensor of the object is calculated based on the RGB and
RGBD images.
(1)
Step
Process
Specifications
Software
1
Image
Capture
-
C#-
Kinect
2
HSL
Filtering
Hue: 60 – 15
Saturation: 0.5 - 1
Luminance: 0-1
C#-
AForge
3
Gray Scale
-
C#-
AForge
4
Threshold
Threshold value 1
C#-
AForge
5
Erosion
x 1 (one time)
C#-
AForge
6
Dilatation
x 5 (five times)
C#-
AForge
7
Centroid
Detection
and labeling
-
Matlab
Cáceres, C.A., Rosário, J.M., Amaya, D.
Copyright © 2016 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol. xx, n.
x
II.2. Brain computer Interface System
The Brain Computer Interface (BCI) system allows the
user to interact with the system. The used method requires
an EEG sensor and a visual stimulus source. The used
method is the SSVEP based on the method proposed by
[24] and adjusted to the Emotiv EPOC sensor by [25]. The
method was adapted to the architecture, using the PC LCD
screen (60 Hz), for the stimulus, and the Emotiv EPOC, as
the EEG sensor.
The generated visual stimuli is created by a developed
software, coded in Processing 2.2.1, a based Java language
compiler. The user has on the screen 3 blinking buttons, 3
different colored squares with a number inside, which
blink in different frequencies.
The buttons are distributed in the screen as is shown in
Fig 3; each button represents the object or objects detected
by the Kinect Camera in the interest area.
Fig 3. Stimulus Software Developed
The buttons have different flashing frequencies and
colors, chosen according to [24]–[28] and the physical
limitations of the available hardware, like the screen
frequency update (60 Hz) and the sample frequency of the
Emotiv EPOC EEG (128 Hz). The chosen colors and
frequencies are shown in Table 2.
TABLE II
SSVEP STIMULUS SOFTWARE SPECIFICATIONS
Button
Flickering
Frequency
Color (HEX)
1
6 Hz
Red (#00BDFF)
2
4.3 Hz
Blue (#00BDFF)
3
5 Hz
Purple (#E500FF)
According to the procedure, the next step is to capture
and analyze the EEG stimulated data. In this case, the data
capture and processing was made using Matlab, the Signal
Processing Toolbox and the Emotiv System libraries for
capturing the EEG data row.
According to [24] the EEG sensor signal of interest is
Oz, but the adaptation made by [25] showed the possibility
of using O1 and O2 in the case of using the Emotiv EPOC
sensor. The location of the sensors O1 and O2 are shown
Fig 4.
The chosen signal sources EEG locations, O1 and O2,
which are associated with the brain visual cortex and
because of that; they are directly associated with the
chosen BCI method, the SSVEP. The signals processing
procedure and specifications are showed in Table 3, after
some pre-processing.
The goal of the preprocessing is to remove the signal
noise and calculate the Fast Fourier Transform (FFT) and
after the Power Spectral Density (PSD), then acquire the
frequency with the higher power for the PSD in each taken
sample, according to the procedure proposed by [24] and
tested and adjusted by [25] to the Emotiv EPOC.
Fig 4. Location of the positions O1 and O2 for the EEG sensor in the
10-20 System (Image Adapted of [25])
TABLE III
SSVEP SIGNAL PROCESSING
Step
Process
Details
1
Average of the signals O1 and O2
(O1+O2)/2
2
Signal Autocorrelation
-
3
Hanning Window
2 seconds
4
Bandpass Filtering
4.0 - 15 Hz
5
Fast Fourier Transform (FFT)
-
6
Power Spectral Density (PSD)
-
By another side, the chosen signal classification system
is based on the highest power of the frequency added with
the power of the first harmonic in the PSD. The
dominating frequency class will be the one with the
highest power in the PSD, the same rules are proposed by
[24] and [25], with processing shown in Table 3 and the
variable definitions according to [24] and [25]:
Variables:
SData: Most recent two seconds of EEG data.
CData: A concatenation of up to the three most recent
sets of SData.
Cáceres, C.A., Rosário, J.M., Amaya, D.
Copyright © 2016 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol. xx, n.
x
And classification rules according to [24] and [25],
where the analysis is made after the preprocessing and
detection of the highest PSD frequency power according
to the procedure:
The second greatest value in SData < 0.35 (Empirical
value according to [5]).
The second greatest value in CData < 0.45 (Empirical
value according to [5]).
The same class is domination in four consecutive
iterations.
The pseudocode of the last rules is presented in Fig 5,
where the function PSD_Highf returns the associated
group to the highest frequency power in the PSD.
Fig 5. Pseudocode of the Classifier
After classifying the signal and recognizing the interest
object, it is possible to get the object position measured
from the Kinect camera.
II.3. Robotic Arm of 6-DOF
In the present architecture, the chosen robot is a non-
specific 6-DoF robot with spherical wrist, showed in Fig
6. Modeled using the Denavit–Hartenberg (DH)
parameters it is possible to construct the robot Direct
Kinematics. The table 4 shows the DH parameters of the
chosen robot, where
and . The software used
for the simulation of the robot arm is Matlab and the
Robotics Toolbox [29].
Fig 6. 6-DoF Robotic Arm. Image made using the RoboAnalyzer
Academic Free Software (Available on: http://roboanalyzer.com/ ).
TABLE IV
DH PARAMETERS
Link
1
2
3
4
5
6
For calculating the inverse kinematic, the kinematic
decoupling method was used according to [30], and for the
trajectory planning the polynomial interpolation method
in the joints space proposed in [31] was used. The
corresponding equations with the 5th order polynomial
interpolation for the trajectory planning are presented in
(2) and (3), with zero initial conditions for two steps.
(2)
(3)
Cáceres, C.A., Rosário, J.M., Amaya, D.
Copyright © 2016 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol. xx, n.
x
II.4. Integration
Retaking the manipulator initial point could be a default
point. By another way, the interest point position was
chosen by the user by using the BCI then measured with
the Kinect. Finally, the final point is chosen depending on
the robot task, which could be place a picked object, hold
a bottle of water for the user, or even feed the patient.
It is important to transform the position of the interest
object measured into coordinates of the robot frame. In
this case, the available information is the Kinect location
and the object position measured from the Kinect frame,
then using homogeneous transformations matrices, it is
possible to find the object position from the robot frame
as shown in (4).
(4)
The used system is based on Fig 7, where Pi is the
interest point, So the robot frame and Kinect frame will be
named as k, and, with this configuration and frames
position it is possible to modify the equation (4) and using
the principle of homogeneous transformation matrix to
join a rotation and translation and obtain (5).
Fig 7. Used Simulation
(5)
Now with the position of the initial default point, the
interest point found using (5) and the final point, it is
possible to create a trajectory using the inverse kinematics
based on table 4 and the equation (4). The Fig 8 shows the
trajectories between the desired points and the full
environment simulated in Matlab.
Fig 8. Simulated Environment
III. Results
Through the proposed approach to assistive robotics, it
is possible to understand the integration process and the
involved processing of each system. The mathematics and
engineering processes involved in the proposed approach
are presented in blocks in Fig 9.
Fig 10. Blocks diagram of the Proposed Architecture
Aiming to the integration results as a full-integrated
system is important to check the most results of each
subsystem to determine the global result.
The first subsystem to analyze is the image processing
integrated with the BCI software. By one side, the objects
detection through image processing is accurate for the
given laboratory conditions with controlled light and
external noise. The object selection depends entirely on
the brain signal processing and a good screen flickering.
The next system to analyze is the BCI SSVEP system
performance, the time response, classifier accuracy and
the ITR (information transfer rate) [32], the results are
presented in table 5. According to the table, it is possible
Cáceres, C.A., Rosário, J.M., Amaya, D.
Copyright © 2016 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol. xx, n.
x
to compare and check the similar results with the tests
made by [24] and [25], and recognize the accuracy and
ITR, and acceptable selection time for each choice.
TABLE V
SSVEP BCI PERFORMANCE RESULTS
Test
Total
Selections
Average
Selection
time (s)
Accuracy
(%)
ITR
(bits/min)
1
57
9.50
83.33
33.14
2
49
8.17
83.37
36.95
3
25
8.33
100
33.44
4
22
11.00
100
24.32
5
30
7.50
100
39.26
6
46
7.66
83.26
38.66
Mean ±
Std
38.17 ±
14.39
8.69 ±
1.33
91.65 ±
9.13
34.30 ±
5.52
It is important to remember, the object selection for the
proposed architecture needs two bits, then according to the
table 5, the needed time for choosing an object is 8,69 ±
1.33 seconds, without counting the relaxing time or any
another time; with an accuracy of 91.65 ± 9.13 %. The Fig
11 shows the selection of an object in each step.
Fig 11. Object Selection
The last system to analyze is the robotic arm and its
trajectories between the initial (0.595, 0, 0.83) m, interest
(0.3, 0.24, 0.35) m and final point (0.4, -0.6, 0.65) m with
the chosen 5th order polynomial, with the Kinect in the
coordinates (0.3, 0.3, 1.35) m. The behavior of the joints
(q) position (Fig 12); speed (Fig 13) and acceleration (Fig
14). The full 3D trajectory is presented in Fig 15, and the
full simulation system was already presented in Fig 8.All
the 3D simulations present an offset in the Z-axis of -0.35.
Fig 12. Joints Position
Fig 13. Joints Speed
Fig 14. Joints Acceleration
Cáceres, C.A., Rosário, J.M., Amaya, D.
Copyright © 2016 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol. xx, n.
x
Fig 15. Robot Arm Trajectory
IV. Conclusions
It can be concluded that the proposed Architecture for
Assistive Robotics is a reality with the current technology,
whereas the correct integration and synergy of the
different branches of the engineering and applied
mathematics is very important. The Assistive Robotics is
a need to fulfill in the modern society and provides an
opportunity for improving the quality of life, especially in
assisting elder and disabled people in daily tasks, by tasks
such as objects pick and place, eat or drink assistance,
displacement, etc.
In addition, the proposed architecture is a valid
methodology for a robotics assistance architecture and
could be comparable to the human way of carrying out a
basic daily task, where the human decides and acts. The
architecture starts with sensing the interest area, by what
could be the human vision system. The next step is the
human choice of choosing the desired object using a BCI
system, which could be comparable to a brain decision.
The last task is planning and executing the assistive task
with a robotic system, which could be comparable with the
movement and action of the human hands, arms or legs to
achieve a goal.
Finally, the integration of the different fields of
engineering aim to explore new and interesting areas for
improving human quality of life like the studied area in
this article, the assistive robotics.
Acknowledgements
This work was supported by the National Council for
the Improvement of Higher Education (CAPES) with a
scholarship for Master Studies in Mechanical Engineering
at the State University of Campinas (UNICAMP).
References
[1] Centers for Disease Control and Prevention (CDC), Ten great
public health achievements--worldwide, 2001-2010., vol. 60,
no. 24. American Medical Association, 2011.
[2] H. Lancaster, “Mortality and the Evolution of Society,”
Expect. Life, pp. 1–11, 1990.
[3] R. Ortner, B. Z. Allison, G. Korisek, H. Gaggl, and G.
Pfurtscheller, “An SSVEP BCI to control a hand orthosis for
persons with tetraplegia,” IEEE Trans. Neural Syst. Rehabil.
Eng., vol. 19, no. 1, pp. 1–5, Feb. 2011.
[4] G. Pfurtscheller, T. Solis-Escalante, R. Ortner, P. Linortner,
and G. R. Muller-Putz, “Self-paced operation of an SSVEP-
based orthosis with and without an imagery-based ‘brain
switch’: A feasibility study towards a hybrid BCI,” IEEE
Trans. Neural Syst. Rehabil. Eng., vol. 18, no. 4, pp. 409–
414, 2010.
[5] L. Montesano, M. Díaz, S. Bhaskar, and J. Minguez,
“Towards an intelligent wheelchair system for users with
cerebral palsy.,” IEEE Trans. Neural Syst. Rehabil. Eng., vol.
18, no. 2, pp. 193–202, Apr. 2010.
[6] Z. Bahri, S. Abdulaal, and M. Buallay, “Sub-band-power-
based efficient Brain Computer Interface for wheelchair
control,” in 2014 World Symposium on Computer
Applications and Research, WSCAR 2014, 2014, vol. 4, no. 1,
pp. 34–40.
[7] R. Leeb, D. Friedman, G. R. Müller-Putz, R. Scherer, M.
Slater, and G. Pfurtscheller, “Self-paced (asynchronous) BCI
control of a wheelchair in virtual environments: A case study
with a tetraplegic,” Comput. Intell. Neurosci., vol. 2007,
2007.
[8] B. Rebsamen, E. Burdet, C. Guan, H. Zhang, C. L. Teo, Q.
Zeng, C. Laugier, and M. H. Ang Jr., “Controlling a
Wheelchair Indoors Using Thought,” IEEE Intell. Syst., vol.
22, no. 2, pp. 18–24, Mar. 2007.
[9] E. J. Rechy-Ramirez, H. Hu, and K. McDonald-Maier, “Head
movements based control of an intelligent wheelchair in an
indoor environment,” 2012 IEEE Int. Conf. Robot.
Biomimetics, pp. 1464–1469, 2012.
[10] S. M. T. M??ller, T. F. Bastos, and M. S. Filho, “Proposal of
a SSVEP-BCI to command a robotic wheelchair,” J. Control.
Autom. Electr. Syst., vol. 24, no. 1–2, pp. 97–105, Jun. 2013.
[11] M. Palankar, K. J. De Laurentis, R. Alqasemi, E. Veras, R.
Dubey, Y. Arbel, and E. Donchin, “Control of a 9-DoF
wheelchair-mounted robotic arm system using a P300 brain
computer interface: Initial experiments,” in 2008 IEEE
International Conference on Robotics and Biomimetics,
ROBIO 2008, 2008, pp. 348–353.
[12] S. Fok, R. Schwartz, M. Wronkiewicz, C. Holmes, J. Zhang,
T. Somers, D. Bundy, and E. Leuthardt, “An EEG-based
brain computer interface for rehabilitation and restoration of
hand control following stroke using ipsilateral cortical
physiology,” Proc. Annu. Int. Conf. IEEE Eng. Med. Biol.
Soc. EMBS, pp. 6277–6280, 2011.
[13] J. Webb, Z. G. Xiao, K. P. Aschenbrenner, G. Herrnstadt, and
C. Menon, “Towards a portable assistive arm exoskeleton for
stroke patient rehabilitation controlled through a brain
computer interface,” Proc. IEEE RAS EMBS Int. Conf.
Biomed. Robot. Biomechatronics, pp. 1299–1304, 2012.
[14] T. Lüth, D. Ojdanić, O. Friman, O. Prenzel, and A. Gräser,
“Low level control in a semi-autonomous rehabilitation
robotic system via a brain-computer interface,” in 2007 IEEE
10th International Conference on Rehabilitation Robotics,
ICORR’07, 2007, pp. 721–728.
[15] A. Vourvopoulos and F. Liarokapis, “Brain-controlled NXT
Robot: Tele-operating a robot through brain electrical
activity,” Proc. - 2011 3rd Int. Conf. Games Virtual Worlds
Serious Appl. VS-Games 2011, pp. 140–143, 2011.
[16] A. Güneysu and H. L. Akin, “An SSVEP based BCI to
control a humanoid robot by using portable EEG device,”
Conf. Proc. ... Annu. Int. Conf. IEEE Eng. Med. Biol. Soc.
IEEE Eng. Med. Biol. Soc. Annu. Conf., vol. 2013, pp. 6905–
6908, Jul. 2013.
[17] C. Escolano, J. M. Antelis, and J. Minguez, “A telepresence
Cáceres, C.A., Rosário, J.M., Amaya, D.
Copyright © 2016 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol. xx, n.
x
mobile robot controlled with a noninvasive brain-computer
interface.,” IEEE Trans. Syst. Man. Cybern. B. Cybern., vol.
42, no. 3, pp. 793–804, Jun. 2012.
[18] P. Kumari and A. Vaish, “Brainwave based user
identification system: A pilot study in robotics environment,”
Rob. Auton. Syst., vol. 65, pp. 15–23, 2015.
[19] A. C. Villa-Parra, D. Delisle-Rodríguez, A. López-Delis, T.
Bastos-Filho, R. Sagaró, and A. Frizera-Neto, “Towards a
robotic knee exoskeleton control based on human motion
intention through EEG and sEMGsignals,” Teodiano Bastos-
Filho/ Procedia Manuf., vol. 00, no. Ahfe, pp. 0–0, 2015.
[20] C. A. Cáceres Flórez, O. L. Ramos Sandoval, and D. Amaya
Hurtado, “Procesamiento de imágenes para reconocimiento
de daños causados por plagas en el cultivo de Begonia
semperflorens (flor de azúcar),” Acta Agronómica, vol. 64,
no. 3, pp. 273–279, 2015.
[21] G. Louverdis, M. I. Vardavoulia, I. Andreadis, and P.
Tsalides, “A new approach to morphological color image
processing,” Pattern Recognit., vol. 35, no. 8, pp. 1733–1741,
2002.
[22] J. Smisek, M. Jancosek, and T. Pajdla, “3D with Kinect,” in
Consumer Depth Cameras for Computer Vision, London:
Springer London, 2011, pp. 3–25.
[23] J. Figueroa and L. Contreras, “Development of an Object
Recognition and Location System Using the Microsoft Kinect
TM Sensor,” Rob. 2011 Robot …, pp. 440–449, 2012.
[24] A. Vilic, T. W. Kjaer, C. E. Thomsen, S. Puthusserypady, and
H. B. D. Sorensen, “DTU BCI speller: An SSVEP-based
spelling system with dictionary support,” in Proceedings of
the Annual International Conference of the IEEE Engineering
in Medicine and Biology Society, EMBS, 2013, pp. 2212–
2215.
[25] F. T. Hvaring and A. H. Ulltveit-moe, “A Comparison of
Visual Evoked Potential ( VEP ) -Based Methods for the
Low-Cost Emotiv EPOC Neuroheadset,” Norwegian
University of Science and Technology, 2014.
[26] R. J. M. Godinez Tello, S. M. T. M??ller, A. Ferreira, and T.
F. Bastos, “Comparison of the influence of stimuli color on
steady-state visual evoked potentials,” Rev. Bras. Eng.
Biomed., vol. 31, no. 3, pp. 218–231, Sep. 2015.
[27] T. Cao, F. Wan, P. U. Mak, P. I. Mak, M. I. Vai, and Y. Hu,
“Flashing color on the performance of SSVEP-based brain-
computer interfaces,” in Proceedings of the Annual
International Conference of the IEEE Engineering in
Medicine and Biology Society, EMBS, 2012, pp. 1819–1822.
[28] M. Aljshamee, M. Q. Mohammed, R.-U.-A. Choudhury, A.
Malekpour, and P. Luksch, “Beyond Pure Frequency and
Phases Exploiting: Color Influence in SSVEP Based on BCI,”
Comput. Technol. Appl., vol. 5, pp. 111–118, 2014.
[29] P. Corke, Robotics, vision and control: fundamental
algorithms in MATLAB. Springer Science & Business Media,
2011.
[30] R. N. Jazar, “Inverse Kinematics,” Theory Appl. Robot., pp.
263–296, 2007.
[31] Y. Guan, K. Yokoi, O. Stasse, and A. Kheddar, “On robotic
trajectory planning using polynomial interpolations,” 2005
IEEE Int. Conf. Robot. Biomimetics - ROBIO, pp. 111–116,
2005.
[32] W. Speier, C. Arnold, and N. Pouratian, “Evaluating true bci
communication rate through mutual information and
language models,” PLoS One, vol. 8, no. 10, 2013.
Authors’ information
Camilo Andrés Cáceres Studied at UMNG
(Nueva Granada Military University), Bogotá,
Colombia reciving the B Sc. degree in
Mecatronics Engineering in 2014. Right now
is in the last stage of his M. Sc. In Mechanical
Engineering at Campinas State University,
São Paulo, Brazil. His interests are the
technology integration, brain computer
interfaces, signal and image processing,
software development, control and systems modeling, robotics and
optimization based in Meta-heuristics and bio-inspired methods.
João Maurício Rosário graduated in
Mechanical Engineer at the Universidade
Estadual de Campinas (1981), he hold a
Master in Industrial Automatic Production
from the Universite de Nancy I (1986), master
in Mechanical Engineering from Universidade
Estadual de Campinas (1984) and a PhD in
Automation, Specialist in Robotics from the
École Centrale des Arts et Manufactures Paris
(1990). He has experience in Mechatronics engineering, acting on the
following subjects: Robotics, Modelling and Control, and Industrial
Automation.
Dario Amaya was educated at UAN, Bogotá,
Colombia receiving the B Sc. degree in
Electronics Engineering in 1995 and the M.Sc.
degree in Teleinformatic in 2007 by the
Faculty of Engineering at the Francisco José
de Caldas District University, UFJC in
Bogotá, Colombia. He was awarded the Ph.D.
degree in 2011 in Mechanical Engineering at
Campinas State University, São Paulo, Brazil,
working on hybrid control. He had worked as a professor and researcher
at the Military University, Colombia since 2007 been involved in
Robotics, Mechatronics and Automation areas.
