of a Mobile PlatforM iN robotized. Wireless seNsor NetWorks. Sevil Ahmed, Nikola Shakev, Lilyana Milusheva, Andon Topalov. Department of Control Systems, ...
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Neural Net Tracking Control of a Mobile Platform in Robotized Wireless Sensor Networks Sevil Ahmed, Nikola Shakev, Lilyana Milusheva, Andon Topalov Department of Control Systems, Technical University–Sofia, Branch Plovdiv 25 Tsanko Djustabanov St., 4000 Plovdiv, BULGARIA
June 22-24, 2015, Liberec, Czech Republic
Outline
2 | 17
Introduction Prototyping the robotized sensor nodes The neural net trajectory tracking control approach The experiment Future work
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Introduction
3 | 17
What is WSN? Distributed autonomous sensors - nodes Large number of nodes Collective data logging and transmission
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Introduction
4 | 17
Applications
Monitoring environmental parameters Machine health monitoring Industrial process monitoring and control
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
5 | 17
The trend - inclusion of mobile robots into the WSN structure! Provide flexibility with respect to the installation of the network sensors, thereby to allow active (not passive) information gathering If necessary, robots can perform desired or based on real-time observations interaction with the environment
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Prototyping the robotized sensor nodes
6 | 17
Sensor node components: The communication: two independent WiFi communication channels
Local area network MQTT based
communication
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Prototyping the robotized sensor nodes
7 | 17
A. The Nonholonomic Mobile Robot iRobot Create
Open interface Hardware expansion possibility
Built-in sensors Differential-drive
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Prototyping the robotized sensor nodes
8 | 17
B. The Embedded Microprocessor verdex pro™ XL6P COM Hardware:
Sticky interface Netpro-vx FCC WiFi module Software:
ОЕ Linux - Angstrom distribution 2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Prototyping the robotized sensor nodes C. The Sensor Pack
9 | 17
Tiva™ C Series TM4C1294NCPDT LaunchPad WiFi CC3100 Booster Pack TI Sensor Hub
9-axis MEMS motion tracking; 3-axis gyro; 3-axis accelerometer; 3-axis compass; pressure sensor; humidity and ambient temperature sensor; ambient and infrared light sensor; non-contact infrared temperature sensor 2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Prototyping the robotized sensor nodes The mobile sensor node
10 | 17
Tiva™ C Series TM4C1294NCPDT LaunchPad WiFi CC3100 Booster Pack TI Sensor Hub Gumstix verdex stack
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
The neural net trajectory tracking control approach 11 | 17
NN trajectory tracking control structure
vr (t ) =
v1 (t ) + v2 (t ) 2
;
vl (t ) =
v1 (t ) − v2 (t ) 2
The neural net structure
x11 (t ) = e1 (t ) = vd (t ) − v(t ) x12 (t ) = vd (t )
x21 (t ) = e2 (t ) = θ d (t ) − θ (t )
x22 (t ) = de2 (t ) / dt
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
The neural net trajectory tracking control approach 12 | 17
The neural net structure
The neural net structure
x11 (t ) = e1 (t ) = vd (t ) − v(t ) x12 (t ) = vd (t )
J (t ) =
1 (e j (t )) 2 , j = 1, 2. 2
x21 (t ) = e2 (t ) = θ d (t ) − θ (t ) x22 (t ) = de2 (t ) / dt
compound sine activation
unipolar sigmoid function
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
The experiment 13 | 17
Scenario
Detect any light source above the robot during the trajectory tracking performance Exact trajectory tracking Continuous sensor reading 2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
The experiment 14 | 17
A. Trajectory tracking performance
Controller
Learning rate
Velocity controller Orientation angle controller
300 0.75
Neurons in the Initial weights hidden layer 5 10-3÷1 7 10-3÷1
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
The experiment 15 | 17
B. Sensor Reading and Data Broadcasting
MQTT broker through iot.eclipse.org:1883 MQTT broker through iot.eclipse.org:1883
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Future work 16 | 17
Kuka youBot
UAVs
ACKNOWLEDGMENT The authors gratefully acknowledge the financial support provided within the Ministry of Education and Science of Bulgaria Research Fund Project FNI I 02/6.
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
Future work 17 | 17
Thank you for your attention!
2015 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics
