Download presentation

Presentation is loading. Please wait.

Published byRonald Drinkwater Modified about 1 year ago

1
NANOROBOTICS CONTROL FOR BIOMEDICAL APPLICATIONS Presented by SARA YOUSEF SERRY ELSAYED M.Sc. Degree in Scientific Computing Department, Faculty of Computer & Information sciences, Ain Shams University. Prof. Dr. TAHA ALARIF Professor in Computer Science Department, Faculty of Computer & Information sciences, Ain Shams University. Under supervision of Prof. Dr. SAFAA AMIN Associate Professor in Scientific Computing Department, Faculty of Computer & Information Sciences- Ain Shams University. 1

2
Presentation overview IntroductionIntroduction Literature ReviewLiterature Review Optimization and Learning AlgorithmsOptimization and Learning Algorithms Cooperative Control of Swarm Nanorobot Target DetectionCooperative Control of Swarm Nanorobot Target Detection Human Blood Stream EnvironmentHuman Blood Stream Environment Polar Coordinate Obstacle Avoidance AlgorithmPolar Coordinate Obstacle Avoidance Algorithm Control Movement Algorithm for Swarm Nanorobot in Human EnvironmentControl Movement Algorithm for Swarm Nanorobot in Human Environment Cooperative Control Design for Nanorobots in Drug DeliveryCooperative Control Design for Nanorobots in Drug Delivery ConclusionsConclusions Contributions and PublicationsContributions and Publications 2

3
Introduction Cancer therapies are currently limited to surgery, radiation, and chemotherapy. All three methods risk damage to normal tissues or incomplete eradication of the cancer. 3

4
Introduction ( Follow up ) Motivation The severe toxic side effects of anticancer drugs on healthy tissues. Dose reduction, treatment delay, or discontinuance of therapy. Limit the side effects of cancer chemotherapy on healthy organs Strengthen drug efficiency to cancer Eliminate tumors by delivering medications directly to the tumor. 4

5
Introduction ( Follow up ) Objectives are Destroy the tumor via injecting swarm of nanorobots Avoiding the collision with the blood cells. Implementing a optimization algorithm called (1+1) Evolution Strategy (ES) with -1/5th success rule algorithm. Combine PSO and Polar Coordinate Obstacle avoidance algorithms Adopt our proposed movement control algorithm, for pH sensitive nanorobots 5

6
Outlines √ Introduction Literature Review Literature Review Optimization and Learning Algorithms Optimization and Learning Algorithms Cooperative Control of Swarm Nanorobot Target Detection Cooperative Control of Swarm Nanorobot Target Detection Human Blood Stream Environment Human Blood Stream Environment Polar Coordinate Obstacle Avoidance Algorithm Polar Coordinate Obstacle Avoidance Algorithm Control Movement Algorithm for Swarm Nanorobot in Human Environment Control Movement Algorithm for Swarm Nanorobot in Human Environment Cooperative Control Design for Nanorobots in Drug Delivery Cooperative Control Design for Nanorobots in Drug Delivery Conclusions Conclusions Contributions and Publications Contributions and Publications 6

7
Literature review Richard Feynman in 1959. NEMS (Nano Electro Mechanical Systems ). One billionth of a meter(10 -9 ) Nanomedicine Nanorobots Architecture 7

8
Features of nanorobots Size Bio Compatibility Powering Communication Navigation Diffusion Swarms Removing 8

9
Outlines √ Introduction √ Literature Review Optimization and Learning Algorithms Optimization and Learning Algorithms Cooperative Control of Swarm Nanorobot Target Detection Cooperative Control of Swarm Nanorobot Target Detection Human Blood Stream Environment Human Blood Stream Environment Polar Coordinate Obstacle Avoidance Algorithm Polar Coordinate Obstacle Avoidance Algorithm Control Movement Algorithm for Swarm Nanorobot in Human Environment Control Movement Algorithm for Swarm Nanorobot in Human Environment Cooperative Control Design for Nanorobots in Drug Delivery Cooperative Control Design for Nanorobots in Drug Delivery Conclusions Conclusions Contributions and Publications Contributions and Publications 9

10
Optimization and Learning Algorithms variables definition domain. Artificial intelligence (AI). Correlation between optimization and learning Evolutionary Algorithms (EA). They have three main characteristics: Population-based. Fitness-oriented. Variation-driven. 10

11
Evolutionary Algorithms Genetic Algorithm (Holland et al., 1960’s) Bitstrings, mainly crossover, proportionate selection Evolution Strategy (Rechenberg et al., 1960’s) Real values, mainly mutation, truncation selection Evolutionary Programming (Fogel et al., 1960’s) FSMs, mutation only, tournament selection Genetic Programming (Koza, 1990) Trees, mainly crossover, proportionate selection Swarm Intelligence (Beni and Wang,1989 ) (Considered as Advanced EAs) 11

12
Evolutionary Strategy (ES) An individual in an ES is represented as a pair of real vectors, v = (x,σ) Mutation is performed by replacing x by x t+1 = x t + N(0, σ) ( + ), uses parents and creates offspring. ( , ), works by the parents producing offspring (1 + 1), this took a single parent and produced a single offspring. 12

13
(1+1) Evolutionary Strategy In the (1+1) ES’s, the new individual replaced its parent if it had a higher fitness. In addition, (1+1) ES, maintained the same value for σ throughout the duration of the algorithm. 13

14
(1+1) Evolutionary Strategy with 1/5 th Success Rule Rechenberg has proposed the “1/5 success rule.” The ratio, , of successful mutations to all mutations should be 1/5. Increase the variance of the mutation operator if is greater than 1/5. otherwise, decrease it Motivation behind 1/5 rule: Try larger steps Proceed in smaller steps 14

15
(1+1) ES with 1/5 th Success Rule 1. Create a random initial configuration x 0 2. Evaluate fitness function f(x 0 ) 3. For t=1 to n (number of generations) Do a. Produce µ mutations of x t-1 using: xi j=x i t-1 +σ [t]·Ni(0,1 ) b. forall i n, j=1,2,…, µ i. Generate one child x c by the combination of the m mutations using m=randint(1, m ) ii. xi c= xi m, forall i to n c. Evaluate f(x c ) d. Apply comparison to select the best individual x t between x t-1 and x c f. If (t mod n = 0) Then i. If (ps>1/5) Then σ[t]= σ[t-n]/ c ii. Else If (ps<1/5) Then σ [t]= σ[t-n]·c Else If (ps=1/5) Then σ [t]= σ[t-n ] 15

16
Swarm Intelligence (SI) Beni and Wang in 1989 with their study of cellular robotic systems. The concept of SI was expanded by Bonabeau, Dorigo, and Theraulaz in 1999. Two common SI algorithms : Ant Colony Optimization Particle Swarm Optimization 16

17
Particle Swarm Optimization(PSO) Proposed by James Kennedy & Russell Eberhart in 1995 Inspired by social behavior of birds and fishes Combines self-experience with social experience 17

18
Performance of PSO Algorithms Relies on selecting several parameters correctly 18 Constriction factor Used to control the convergence properties of a PSO Inertia weight How much of the velocity should be keeped from previous steps Cognitive parameter The individual’s “best” success so far Social parameter Neighbors’ “best” successes so far Vmax Maximum velocity along any dimension

19
Particle Swarm Optimization Swarm: a set of particles (S) Particle: a potential solution Position: Velocity: Each particle maintains Individual best position (PBest) Swarm maintains its global best (GBest) 19 S Fitness function Fitness value

20
Particle Swarm Optimization(PSO) Pbest. Gbest. The basic concept of PSO lies in accelerating each particle toward its pbest and the gbest locations. 20

21
Particle Swarm Optimization(PSO) Algorithm Basic algorithm of PSO 1.Initialize the swarm form the solution space 2.Evaluate the fitness of each particle 3.Update individual and global bests 4.Update velocity and position of each particle 5.Go to step2, and repeat until termination condition 21

22
PSO and ES Comparison Commonalities Population based optimization. Randomly generated population. Fitness values Update the population Both systems do not guarantee success. Differences PSO does not have genetic operators memory Particles do not die The information sharing mechanism ES population moves together 22

23
Outlines √ Introduction √ Literature Review √ Optimization and Learning Algorithms Cooperative Control of Swarm Nanorobot Target Detection Cooperative Control of Swarm Nanorobot Target Detection Human Blood Stream Environment Human Blood Stream Environment Polar Coordinate Obstacle Avoidance Algorithm Polar Coordinate Obstacle Avoidance Algorithm Control Movement Algorithm for Swarm Nanorobot in Human Environment Control Movement Algorithm for Swarm Nanorobot in Human Environment Cooperative Control Design for Nanorobots in Drug Delivery Cooperative Control Design for Nanorobots in Drug Delivery Conclusions Conclusions Contributions and Publications Contributions and Publications 23

24
Cooperative Control of Swarm Nanorobot Target Detection Communication Between Nanorobots This optimization algorithm runs independently on each nanorobot. Each nanorobot optimizes only its own plan. Nanorobots maintain a record of movement plans 24

25
Mutation Strategies 25 Straight Strategy Swap Strategy High Probability Strategy The nanorobot will set its entire movement plan as a straight line in a random direction Two randomly chosen vectors in the movement plan will be swapped A randomly vectors in the movement plan are rotated by a random angle taken from a normal distribution causing the entire path to be rotated

26
Simulation Results Straight Strategy Swap Strategy High Probability Strategy 26

27
Simulation Analysis The final target areaThe final target area 27 The average time

28
Simulation Analysis ( FOLLOW UP ) The average time in the Partial optimization levelThe average time in the Partial optimization level 28 The average time in the Full optimization level

29
Simulation Analysis ( FOLLOW UP ) The swap strategy is inefficient mutation strategy The straight and high strategies have almost the same number of swarm of nanorobots High strategy is more efficient than straight strategy in the partial and full optimization levels. 29

30
Outlines √ Introduction √ Literature Review √ Optimization and Learning Algorithms √ Cooperative Control of Swarm Nanorobot Target Detection Human Blood Stream Environment Human Blood Stream Environment Polar Coordinate Obstacle Avoidance Algorithm Polar Coordinate Obstacle Avoidance Algorithm Control Movement Algorithm for Swarm Nanorobot in Human Environment Control Movement Algorithm for Swarm Nanorobot in Human Environment Cooperative Control Design for Nanorobots in Drug Delivery Cooperative Control Design for Nanorobots in Drug Delivery Conclusions Conclusions Contributions and Publications Contributions and Publications 30

31
Human Blood Stream Environment In this study we solve the path planning problem of swarm nanorobot The blood cells are obstacles in the nanorobot movement Blood flow, blood viscosity and blood density 31

32
Blood Physical Properties Blood velocity in the pipe ~1mm/sec Blood flow is the actual volume of blood flowing through a vessel, an organ, or the entire circulation at a given time. 32

33
Outlines √ Introduction √ Literature Review √ Optimization and Learning Algorithms √ Cooperative Control of Swarm Nanorobot Target Detection √ Human Blood Stream Environment Polar Coordinate Obstacle Avoidance Algorithm Polar Coordinate Obstacle Avoidance Algorithm Control Design of Swarm Nanorobot in Human Environment Control Design of Swarm Nanorobot in Human Environment Cooperative Control Design for Nanorobots in Drug Delivery Cooperative Control Design for Nanorobots in Drug Delivery Conclusions Conclusions ContributionsPublications Contributions and Publications 33

34
Polar Coordinate Obstacle Avoidance Algorithm The nanorobot have sensors to detect obstacles (blood cells ) Self organized trajectory planning is required to avoid obstacles. 34

35
Polar Coordinate Obstacle Avoidance The new position of obstacle (x j,y j ) within time Δt can be calculated : x j = x i + v f x i * Δt ;y j = y i + v f y i *Δt The distance Δd can be calculated by : 35

36
Outlines √ Introduction √ Literature Review √ Optimization and Learning Algorithms √ Cooperative Control of Swarm Nanorobot Target Detection √ Human Blood Stream Environment √ Polar Coordinate Obstacle Avoidance Algorithm Control Movement Algorithm for Swarm Nanorobot in Human Environment Control Movement Algorithm for Swarm Nanorobot in Human Environment Cooperative Control Design for Nanorobots in Drug Delivery Cooperative Control Design for Nanorobots in Drug Delivery Conclusions Conclusions ContributionsPublications Contributions and Publications 36

37
Control Movement Algorithm for Swarm Nanorobots Global path and local path are considered for nanorobot’s movement path planning. Global path is carried with some modifications in PSO algorithm. When obstacles are encountered, the local path planning is found out for obstacle avoidance. 37

38
Local Path Planning The goodness of the position can be computed by using the fitness function Fj. Fitness function for each nanorobot at k th iteration is represented by: Fj (k) = max F j (s i where s i ∈ S, s i ∉ T obstacle ) 38

39
Global Path Planning Total area covered = The pbest F i will be the best fitness value obtained by a nanorobot at a selected time. F i = E[Di j ] The gbest F g will be the global fitness value of a swarm of neighbor nanorobots at the selected time. Fg = max (F i (N(s i (k)))) 39

40
Global Path Planning The velocity is updated in the k th iteration by using : V i (k+1) = R+ w i v i (k)+ c1 * r1 * (F i (k) – s i (k)) + c 2 * r 2 * (F j (k) – s i (k)) The velocity v of the nanorobot decides where it moves next by using the following equation. si (k+1)=si (k)+vi( k+1 ) 40

41
Movement Control Algorithm 41 The improved PSO algorithm the obstacle avoidance algorithm The improved PSO algorithm.

42
Polar Coordinate Obstacle Avoidance Algorithm Calculate Δd, distance between nanorobot si and obstacle Calculate time to collision Δtc based on Δd If Δd < threshold θ = θ + 180 ○ If Δd > threshold and the target area are in the positive y-axis direction Δθij = Δθij + 90 ○ If Δd > threshold and the target area are in the negative y-axis direction Δθij = Δθij -90 ○ Calculate Cartesian coordinate’s xij,yij from polar coordinates where xij is r*Cos(Δθij); yij is r* Sin(Δθij); r is the radius of the obstacle (0< Δθij<180) Move nanorobot si from xi1,yi1 to xij,yij 42

43
Calculate Δd Calculate time to collision Δtc Calculate Δθij Calculate Cartesian coordinate s Move nanorobot si from xi1,yi1 to xij,yij Calculate coverage of range si to its neighbors N(si) If coverage value>curr ent optimum Current optimum target=curren t selected target Move nanorobot si to new best position End 43 Movement Control Algorithm

44
Simulation Schema C programming environment. Both nanorobot and obstacle flow with same fluid velocity. Nanorobot and obstacle has same radium. Nanorobot has Re ≈ 10−3. We consider a constant velocity and ignore some stochastic factors for simplicity. 44

45
Simulation Parameters Nanorobot radius 3 µm Nanorobot radius 10 µm Red cell radius 7 µm White cell radius 12 µm Blood viscosity 10-2 g/cm.s Blood velocity 100 μm /s Blood density 1 g/cm3 T free 40 s 45

46
Simulation for Swarm of 10 Nanorobots Demonstrates that all the nanorobots reach the target area effectively and in 48.2189 seconds 46

47
Simulation Analysis 47 Percentage of coverage in each time interval Time required for each nanorobot to generate the best value. Coverage conveys the percentage of target cells received by the nanorobots.

48
Comparison between PSO and High Mutation Strategy ES PSO 48 Using a set of benchmark test problems

49
Comparison between PSO and High Mutation Strategy ES PSO 49 The full coverage achieved by all of the nanorobots.

50
Outlines √ Introduction √ Literature Review √ Optimization and Learning Algorithms √ Cooperative Control of Swarm Nanorobot Target Detection √ Human Blood Stream Environment √ Polar Coordinate Obstacle Avoidance Algorithm √ Control Movement Algorithm for Swarm Nanorobot in Human Environment Cooperative Control Design for Nanorobots in Drug Delivery Cooperative Control Design for Nanorobots in Drug Delivery Conclusions Conclusions ContributionsPublications Contributions and Publications 50

51
Cooperative Control Design for Nanorobots in Drug Delivery Existing Control Strategies :- Ishida, design behavior-based source Goodman,the method is extended to scenarios with a group of nanorobots. Gazi, give a control law by combining a potential field control law and a gradient based control law. Zhang,use extremum-seeking control theories 51

52
Existing Control Strategies :- ( Follow up ) Ogren solved the problem by Least square method Bachmayer two strategies, the 1 st for a single robot with historical data and the 2 nd uses a group of robots with projected gradient estimation. Michael, a stochastic gradient-ascent algorithm Grzybowski, notes that cancers are more acidic than the rest of the body Low pH value 52

53
pH Sensitive Nanorobots In a tumor microenvironment, the pH distribution is measurable, volume, concentration, displacement and velocity.Chemical Sensors are used to measure changes in volume, concentration, displacement and velocity. pH sensitive nanorobot is promoted as an alternative treatment for cancer. 53

54
High pH Therapy The 'High pH Therapy‘ prevents cancer cells from undergoing mitosis Anaerobic metabolism Produces lactic acid This alters DNA to allow uncontrolled growth. Also causes pain. 54

55
High pH Therapy ( Follow up ) A swarm of pH sensitive nanorobot: increases the intracellular pH of tumor cells. Generates alkaline solution. Though given as the chloride salt. Follows sodium pathway into cells. Raises intracellular pH to 8 The resulting alkaline environment result in cell death. Ends pain 55

56
Tumor Microenvironment We concentrate on low pH value of the target searching method 56

57
Control Problem We define our control objective to be the group of robots reaching the tumor area, which is defined by certain pH value around the tumor area. 57

58
Control Algorithm for Drug Delivery in tumor The 1 th nanorobot didn't detect an obstacle. It moves according to the PSO algorithm The nanorobot receives pH value, position and velocity of all the other nanorobots by communication using PSO. Until one or more nanorobots has a measurement of the pH value less than 7.0. Consequently the high pH therapy will be applied to destroy the tumor. 58

59
Simulator Platform 59 Representation of tumor pH environment in drug delivery system

60
Simulation Results A group of 25 pH sensitive nanorobots in drug delivery system 60

61
Simulation Analysis 61

62
Outlines √ Introduction √ Literature Review √ Optimization and Learning Algorithms √ Cooperative Control of Swarm Nanorobot Target Detection √ Human Blood Stream Environment √ Polar Coordinate Obstacle Avoidance Algorithm √ Control Movement Algorithm for Swarm Nanorobot in Human Environment √ Cooperative Control Design for Nanorobots in Drug Delivery Conclusions Conclusions Contributions and Publications Contributions and Publications 62

63
Conclusion In this study, we Developed cooperative control strategies Concluded that the high strategy is more efficient than the straight strategy Introduced Behavior-based robot navigation methods 63

64
Conclusion (Follow up) The proposed scheme effectively constructs an obstacle free self-organized trajectory. The simulation results constructed an obstacle free self- organized path. 64

65
Conclusion (Follow up) We designed control strategies for nanorobots to Trace the gradient of the measured pH values Reach the tumor cells with the lowest pH value. Also, the capability of the control strategy is illustrated through simulating a scenario of drug delivery by a group of nanorobots. 65

66
Outlines √ Introduction √ Literature Review √ Optimization and Learning Algorithms √ Cooperative Control of Swarm Nanorobot Target Detection √ Human Blood Stream Environment √ Polar Coordinate Obstacle Avoidance Algorithm √ Control Movement Algorithm for Swarm Nanorobot in Human Environment √ Cooperative Control Design for Nanorobots in Drug Delivery √ Conclusions Contributions and Publications Contributions and Publications 66

67
Contributions Improving the (1+1) evolutionary strategy with 1/5 th success rule algorithm Comparing between the three mutation strategies Improving the PSO algorithm for the purpose of communication between nanorobots. 67

68
Contributions (Follow up) Modifying the obstacle avoidance algorithm to enable nanorobot to avoid blood cell. Studying the effects of the fluid flow of the blood on the motion of nanorobots. 68

69
Contributions (Follow up) Combining PSO and obstacle avoidance algorithms to control nanorobots’ behavior. Developing a new control algorithm for pH sensitive nanorobots to Simulating the pH tumor environment and the process of nanorobots in the drug delivery system. 69

70
References and Publications ] S.Ahmed, S.E. Amin, T. Alarif,“A Novel Communication Technique for Nanorobots Swarms Based on Evolutionary Strategies”, Proceedings of the UKSim-AMSS 16th International Conference on Computer Modeling and Simulation. [2] S.Ahmed, S.E. Amin, T. Alarif,” Simulation for the Motion of Nanorobots in Human Blood Stream Environment”, Proceedings of the ACV-international Conference on Advances in Computer Vision. [3] S.Ahmed, S.E. Amin, T. Alarif, “Efficient Cooperative Control System for pH Sensitive Nanorobots in Drug Delivery”, International Journal of Computer Applications( IJCA). [4] S.Ahmed, S.E. Amin, T. Alarif, “Assessment of Applying Path Planning Technique to Nanorobots in a Human Blood Environment”, Proceedings of the 2014 UKSim-AMSS 8th European Modeling Symposium on Mathematical Modeling and Computer Simulation. [5] S.Ahmed, S.E. Amin, T. Alarif, “Investigation of Mutation Evolutionary Strategies Applied to Nanorobots”, International Journal of Advanced Robotic Systems, (SUBMITTED). [6] S.Ahmed, S.E. Amin, T. Alarif, “Swarm Nanorobot Path Planning in a Human Blood Environment”, Pattern Recognition Letters (SUBMITTED). 70

71
SARA YOUSEF SERRY ELSAYED 71

Similar presentations

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google