Presentation is loading. Please wait.

Presentation is loading. Please wait.

Fuzzy Logic Applications

Similar presentations


Presentation on theme: "Fuzzy Logic Applications"— Presentation transcript:

1 Fuzzy Logic Applications
Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

2 “Formed the philosophical idea of shades of grey within truth”
Introduction and Fuzzy Logic Origins The idea of Fuzzy Logic is quite recent, right? NO! - its origins date back to at least the time of Greek Philosophers Maybe even China and India - no one exactly knows However, Lotfi Zadeh published the first literature on ‘Fuzzy Sets’ “Formed the philosophical idea of shades of grey within truth” Our Applications Traffic Coordinated Systems and Optimisation Using Fuzzy Logic Single Traditional Intersection - Lewis Network Intersections with Adaptive Systems (Supervisory Controller) - Ron Robotics and Navigational Systems Using Fuzzy Logic Vertical Pendulum and Image Processing - Alasdair Behavioural Based Navigational Systems - Barnaby CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

3 Modelling a Single Intersection
2 arms – North/South and East/West Arrival of vehicles treated as random and uniformly distributed Every second, generate a random number and compare to a fixed value, if the generated number is greater, a vehicle arrives during this time interval Vehicles leave the queue at a light at a constant rate No turning traffic CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

4 Aim Control the duration of the green period and red period on each arm such that the average delay time per vehicle is minimized Delay at any point in time depends on: The arrival rate of cars The number of cars queued at the arm of the intersection The rate at which cars clear the intersection on a green light (assumed fixed) CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

5 Approach Detection pads a sufficient distance before the intersection
Information about vehicle arrival Controller intervenes every 10 seconds during the green period, starting 7 seconds into the period Decide based on the input parameters how long to extend the green period for CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

6 The Fuzzy Approach Fuzzy heuristics allowing linguistic control statements 4 Fuzzy Variables and their fuzzy sets T = time. “very short”, “short”, “medium” etc. A = cars arriving. “many”, “more than a few” etc. Q = cars in the queue. “any”, “less than small” etc. E = extension to green period. Same groups as T Determine E based on T, A and Q CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

7 Membership Functions CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

8 More/Less than operators
CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

9 The Fuzzy Algorithm For each 10 second interval, the extension is determined based on a series of fuzzy control statements. The length of the extension is then determined by defuzzifying the output variable E. The membership functions act as the degree of confidence with which a particular rule can be applied. Control for the first intervention CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

10 Operations CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

11 Making a Control Decision
CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

12 Results The average delay per vehicle was calculated in a series of simulations with varying flow rates on the N/S and E/W arms. Delays were compared to that of a Vehicle Actuated Controller The Fuzzy Controller outperformed the Vehicle Actuated controller in all cases, reducing the average delay per vehicle by around 15% CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

13 Networks of Intersections - Our Problem
We saw how it worked for one single intersection. Now what about a network? Aim: Same - except we want to eliminate congestion building at any single intersection. Additionally create offset adjustment. Approach: Adjust the offset time by extending greens for an optimal amount of time. OPTION A: Individual Fuzzy Controllers Utilising only an upstream intersection OPTION B: Supervisory Fuzzy Controller Analysing traffic volume at the entirety of the network Image Source: Google Maps, 2017 CITS Artificial Intelligence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

14 Networks of Intersections - The Example
The intersections we are analysing are: Stirling Highway x Broadway Stirling Highway x Winthrop Avenue Mounts Bay Road x Hackett Drive Each intersection has a fuzzy logic controller adjusting the green phase depending on the rules. Let’s add a supervisory fuzzy logic controller to adjust for offset based on traffic volume. C B A Image Source: Nearmap, 2017 CITS Artificial Intelligence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

15 Networks of Intersections - Considerations
Direction of Travel Evaluate the traffic volume, speed and distance Find the difference of volume and calculate the appropriate offset times Decide the extension to be made to increase green phase based on traffic volume at each node C B A CITS Artificial Intelligence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

16 Networks of Intersections - Membership Functions
Calculations and fuzzy sets with rules (inputs and outputs) Divided into 3 Fuzzy Sets Based on the rules shown in the single intersection An overall supervisory controller is more appropriate as it analyses volume at all nodes INPUT VARIABLES: Volume Difference 1: VDIFF1 = VEC - ( avg (VSA,VNA)) Volume Difference 2: VDIFF2 = ( maximum (VWA, VEC) ) - VSB [Morning/Afternoon Peak] Volume Difference 3: VDIFF3 = VWA - VNC VSB VWC VEC C B VSA VEB VEB VNC A VWA VEA VNA CITS Artificial Intelligence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

17 Networks of Intersections - Membership Functions
OUTPUT VARIABLES: Extensions to be made to the green phase of the traffic along Eastbound/Westbound, in our case, Stirling Highway. Extension A - Length of Green Extension B - Length of Green Extension C - Length of Green Membership Function: VSB VWC VEC C B VSA VEB VEB VNC A VWA VEA VNA Image Source: Nainar, I. (1996) CITS Artificial Intelligence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

18 Main Roads WA and the use of SCATS
With the assistance of SCATS and traffic data Main Roads WA was able to optimise multiple major arterial routes within the city that cause costly delays. Over 850 traffic signals are controlled by SCATS. Morning peaks Orrong Road: W/b improved by >1 min (7%). Roe Highway: S/b improved by >1 min (7%) while traffic volumes increased by 11%. South Street: W/b improved by ~2 min (9%) while traffic volumes increased by 4%. Afternoon peaks Ennis Avenue/Mandurah Road: W/b improved by ~2 min (9%) - traffic volumes increased 1%. Great Eastern Highway (Farrall Road to Kalamunda Road): E/b improved by >2 minutes (14%) while traffic volumes decreased by 2%. Stirling Highway: E/b and W/b journey times improved by ~1 minute each (5%) while traffic volumes increased by 3-4%. Image Source: Main Roads Western Australia, 2017 SCATS is installed at ~42,000 intersections in >154 cities in 25 countries. Many other Coordinated and Adaptive Traffic Systems are used widely. CITS Artificial Intelligence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

19 Fuzzy Logic Control in Autonomous Robotics
Autonomous robots are required to deal with environments where fuzzy logic can be useful For instance, a robot navigating through a canyon may find it useful to have concepts of “near” and “far” in relation to its distance too and from the cliff face Allows robot to not change direction as frequently as long as it stays within the rough boundaries of “near” and “far” As an example, the robot in the image should not keep correcting its course too frequently as this will drain its battery faster

20 Balancing Vertical Pendulum Robot
Robot must go repeatedly back and forth to balance a vertical pendulum Must learn when to change course and how fast it should go Fuzzy logic allows the robot to take in account inaccuracies in control of it’s speed and perception of its environment

21 Fuzzy logic controller for inverted pendulum Example (S.N. Deepa)
The angular position and the angular velocity of the pendulum are inputs to the fuzzy controller, outputs cart velocity Seven linguistic variables: negative large (NL), negative medium (NM), negative small (NS), zero (ZE), positive small (PS), positive medium (PM) and positive large (PL) Fig. Control rules by S.N.Deepa [1]

22 Example continued (S.N. Deepa)
Fig. Fuzzy Control surface Fig. Fuzzy Controller

23 Fuzzy Logic Computer Vision
Many robots use image recognition to perceive their environments. A poorly implemented image recognition program can easily suck up battery life and exhaust limited CPU and memory supplies.

24 Fuzzy Logic Example (computer vision)
Recognise door frames by recognising the horizontal and vertical lines. Linguistic variables applied to the size, how vertical and horizontal the lines are.

25 Real world examples of robots using fuzzy logic
Real world examples of robots which may (or may not) use fuzzy logic control systems: Boston dynamics Spot, Big Dog, ect. NASA Mars rover Bomb disposal robots Cutting edge robot projects are sadly often not open source and therefore details of how exactly these robots operate are not known by the general public

26 Navigation systems CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

27 Input data Sensor data received by a robot navigating a space has a degree of uncertainty associated with it. This combined with a varying environment make it hard to get reliable data about the exact location of obstacles for deciding a route for a robot. CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

28 Fuzzy behavior based navigation
The uncertainty in the environment makes it difficult precisely. If we use a fuzzy control system we can achieve a solution that approaches an optimum but can be reached in a shorter time as the exact location and type of the obstacle is not needed. CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

29 Example rules The data required for the system is the distance to obstacles in given directions and the position of the end goal of the robot. From this the direction and speed of the robot can be determined. If Front left is Near And Front right is Far, Then Steering is Right If Front left is Far And Front right is Near, Then Velocity is Zero CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

30 Results The fuzzy rules applied allow the robot to navigate spaces efficiently without getting stuck in dead ends. Along with moving avoiding obstacles along the way. CITS Artificial Intelegence and Adaptive Systems The Applications of Fuzzy Logic Alasdair Penman, Barnaby Robertson Hurst, Lewis Tolonen, Ronald Harwood

31 References Pappis, C.P., Mamdani, E.H. (1977) “A Fuzzy Logic Controller for a Traffic Junction”, IEEE Transactions Systems, Man and Cybernetics, 7(10) Nainar, I. (1996). “An Adaptive Fuzzy Logic Controller for Intelligent Networking and Control”, Edith Cowan University Research, Retrieved from Main Roads WA, (2017). “Traffic Signals”, Retrieved from Amur S. Al Yahmedi and Muhammed A. Fatmi (2011) “Fuzzy Logic Based Navigation of Mobile Robots”


Download ppt "Fuzzy Logic Applications"

Similar presentations


Ads by Google