2 i C 2 i Research on Research on C Intelligent Transportation - PowerPoint PPT Presentation

2 i C 2 i Research on Research on C Intelligent Transportation Intelligent Transportation Michel Pasquier Michel Pasquier 2 i C 2 i ) ( C Centre for Computational Intelligence ( ) Centre for Computational Intelligence Nanyang Technological University Nanyang Technological University School of Computer Engineering, Blk N4 #2A- -32, 32, School of Computer Engineering, Blk N4 #2A Nanyang Avenue, Singapore 639798. Nanyang Avenue, Singapore 639798. E- E -mail: pasquier@pmail.ntu.edu.sg mail: pasquier@pmail.ntu.edu.sg http://www.c2i.ntu.edu.sg/ http://www.c2i.ntu.edu.sg/

Overview Overview 2 i C 2 � Research at Research at C i � � Hybrid neuro Hybrid neuro- -cognitive systems cognitive systems � � GenSoYager fuzzy neural network GenSoYager fuzzy neural network � � Applications related to transportation Applications related to transportation � � Intelligent vehicles and driver modelling Intelligent vehicles and driver modelling � � Automated driver training methodology Automated driver training methodology � � Examples of acquired driving skills Examples of acquired driving skills � � Conclusion and future work Conclusion and future work � C 2 i Centre for Computational Intelligence

2 i C 2 Overview of C i Overview of � Personnel Personnel � – 10 faculty members, 1 RF, 2 technicians M- M -CMA CMAC C – 20 PhD and 7 MEng students Associative Associative Memory Memory – 9 PhD and 13 MEng completed � Projects Projects � – 8 funded projects completed (AcRF, NSTB) – 7 collaborative / industrial projects Intelligent Well Log Analysis Intelligent Well Log Analysis Hybrid CI Automated Parking System Hybrid CI Automated Parking System – various other projects Model X% Y% Supervisor + Adaptive Controller – Primary + Plant Controller Input Effort Output – Neuro- -fuzzy Integrated Process Supervision fuzzy Integrated Process Supervision Neuro C 2 i Centre for Computational Intelligence

From Cognitive Science to IS From Cognitive Science to IS � Cognitive Science is Cognitive Science is � – The study of the mind i.e., cognitive processes and their relation to intelligent behavior, learning, perception, language, emotion, etc. – The key to future intelligent systems, humanized technologies and applications – An important, emerging, multidisciplinary field, fast developing in EU, Japan, USA. C 2 i Centre for Computational Intelligence

2 i C 2 Research at C i Research at � Centre for Centre for Computational Intelligence Computational Intelligence Cognitive Informatics Cognitive Informatics � � Adaptive and autonomous systems Adaptive and autonomous systems � � Nature Nature- -inspired systems inspired systems � � Neuro Neuro- -cognitive architectures cognitive architectures � � Decision support systems Decision support systems � – Synthesize human reasoning capabilities and tolerance to uncertain / incomplete information – Applications in robotics, transportation, HCI, manufacturing, medical, finance, education… C 2 i Centre for Computational Intelligence

Building Intelligent Systems Building Intelligent Systems � From From � Machine Intelligence Machine Intelligence AI AI Artificial Intelligence Computational Intelligence Artificial Intelligence Computational Intelligence Fuzzy to to Computation Logical Probabilistic adaptation Quantum CI CI Reasoning Reasoning Computing to Machine to nature- Planning Learning inspired … … Neural Search Computation MI MI Evolutionary l a Computation i t n e l u e DNA q l l e a s r a Computing p C 2 i Centre for Computational Intelligence

Neuro- -Cognitive Architectures Cognitive Architectures Neuro � Convergence of Cognitive Convergence of Cognitive � and Neuro- - sciences sciences and Neuro � Study of the human mind Study of the human mind � � new science for IS new science for IS � – Sensing, diagnosis, guidance – Semantic learning memory – Skill acquisition e.g. driving “Example is the way to learn. “ Example is the way to learn. Example is the only way to learn.” ” Example is the only way to learn. -- Albert Einstein Albert Einstein -- � Science Science � � humanized IS humanized IS � C 2 i Centre for Computational Intelligence

Hybrid Fuzzy Neural Systems Hybrid Fuzzy Neural Systems � Fuzzy Systems Fuzzy Systems � Fuzzy Neural Fuzzy Neural Fuzzy Neural – Pros: rules intuitive and easily comprehended, Networks (FNN) Networks Networks (FNN) (FNN) emulates the human decision process, prior expert knowledge can be easily incorporated combine: combine: combine: the advantages of FS – Cons: manual design of fuzzy sets and rules, the advantages of FS the advantages of FS heuristic tuning of system parameters the capabilities of NN the capabilities of NN the capabilities of NN � Neural Network Neural Networks s � and other techniques and other techniques and other techniques (clustering, tuning, (clustering, tuning, (clustering, tuning, – Pros: self-organised learning / modelling, evolutionary… …) ) evolutionary evolutionary…) fault tolerance, distributed knowledge – Cons: opaque, no prior knowledge, stability and plasticity dilemma, convergence C 2 i Centre for Computational Intelligence

Example: the GenSoYager FNN Example: the GenSoYager FNN � Generic Generic : not not application application- -specific specific : � (in control, finance, medical, etc.) (in control, finance, medical, etc.) � Self Self- -Organising Organising : automated fuzzy set design : automated fuzzy set design � and data clustering � DIC and data clustering DIC � Yager Yager : sound, most accurate : sound, most accurate � inference scheme � Yager inference scheme Yager � Fuzzy Fuzzy (rule (rule- -based) based) : intuitive, understandable : intuitive, understandable � approximate reasoning approximate reasoning � Neural Network Neural Network : rule learning � RuleMap : rule learning � RuleMap + optimization � GA + optimization GA C 2 i Centre for Computational Intelligence

Automated Fuzzy System Design Automated Fuzzy System Design Inference Scheme and Rule 1: IF x 1 is C … x n is G, then y is K Operators Fuzzy System Construction T-norm: MIN, … etc 1) Generate fuzzy sets to cover the Rule n: IF x 1 is A … x n is I, then y is L T-conorm: input and output spaces � MAX, etc Discrete Incremental Clustering Fuzzy Rule Base 2) Decide the fuzzy inference scheme and defuzzification method � Keller-Yager model Input Output Fuzzy Inference Fuzzifier Defuzzifier 3) Generate fuzzy rules from input- Engine output pairs -> RuleMap learning algorithm / rule formulation phase 4) Select the generated rules to form Defuzzification method: Input x 1 Center of Gravity (CoG) the fuzzy rule base � RuleMap / Mean of Maximum (MoM) parameter learning phase A B C D … Input x n Output y E F G H I J K L C 2 i Centre for Computational Intelligence

GenSoYager- -FNN System FNN System GenSoYager 3. RuleMap Parameter Learning phase 3. RuleMap Parameter Learning phase Defuzzification layer Defuzzification layer Consequent layer Consequent layer 2. RuleMap Rule Formulation phase 2. RuleMap Rule Formulation phase Rule layer Rule layer Antecedent layer Antecedent layer 1. Self-organising / Clustering phase 1. Self-organising / Clustering phase Fuzzification layer Fuzzification layer C 2 i Centre for Computational Intelligence

Evaluation of GenSoYager- -FNN FNN Evaluation of GenSoYager � Classification benchmarks Classification benchmarks � – XOR dilemma Performs equally or Performs equally or Performs equally or 1 0.9 0.8 – Iris classification 0.7 better than previous better than previous better than previous 0.6 0.5 0.4 0.3 FNN architectures: FNN architectures: FNN architectures: – 2-spiral problem 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 – Ionosphere classification Falcon, Falcon- -MLVQ, MLVQ, Falcon, Falcon Falcon, Falcon-MLVQ, Falcon- -ART/MART, ART/MART, Falcon Falcon-ART/MART, � Prediction benchmarks Prediction benchmarks CTE towards PIE towards � Ang Mo Kio ANFIS, GA classifier, ANFIS, GA classifier, ANFIS, GA classifier, Changi Lane 3 POPFNN(CRI/TVR) … … – Box-Jenkins gas furnace POPFNN(CRI/TVR) Lane 2 POPFNN(CRI/TVR) … Lane 1 – Financial data reconstruction Lane 5 Lane 4 Towards Upper Serangoon – Mackey-Glass time series – Traffic flow modelling C 2 i Centre for Computational Intelligence

Applications Applications � Decision Support Systems Decision Support Systems � � Classification Classification � – Forensics/security: handwriting, face identification – Process control: chemical, manufacturing – Medical diagnosis: asthma, cancer, diabetes – Transportation planning, navigation, and control � Prediction Prediction � – Financial forecasting: bank failure, stock et al – Medical prognosis: asthma, surgery – Transportation: traffic flow (macro/micro) C 2 i Centre for Computational Intelligence

Application: Traffic Planning Application: Traffic Planning � Dynamic routing of Dynamic routing of � automated taxi fleet automated taxi fleet – Hybrid ant colony + GA optimization – FPGA realization JAM C 2 i Centre for Computational Intelligence