AN INTRODUCTION TO PATTERN RECOGNITION: STATISTICAL, NEURAL NETWORK, AND SYNTACTIC METHODS

By P. MAKAO

CONTENTS PART I BACKGROUND EMBODIED INTELLIGENCE INTRODUCTION A LONG PAST AND A SHORT HISTORY ARCHITECTURE OF EMBODIED INTELLIGENCE ACTION GENERATION BASED ON LEARNING (L ?A) ACTION GENERATION BASED ON MORPHOLOGY (M ?A) MORPHOLOGICAL CONTROL BASED ON ACTION (A ?M) PERCEPTION-DRIVEN MORPHOLOGY TRANSFORMATION (P ?M) LEARNING-DRIVEN MORPHOLOGY OPTIMIZATION (L ?M) PERCEPTION-DRIVEN ACTION GENERATION (P ?A) ACTION-DRIVEN PERCEPTION IMPROVEMENT (A ?P) ACTION-DRIVEN EMBODIED LEARNING (A ?L) DISCUSSION THE GOOD, BAD AND UGLY OF EMBODIED AI EMBODIED INTELLIGENCE AND ROBOTICS FRONTIERS OF EMBODIED INTELLIGENCE INSPIRATION FROM DISEMBODIED MACHINE LEARNING THE ROLES OF MORPHOLOGY SIM2REAL MULTI-MODAL EMBODIED AI AND LANGUAGE GROUNDING LARGE LANGUAGE MODEL FOR EMBODIED AI MULTI-AGENT COLLABORATION SECURITY AND VULNERABILITY OF EMBODIED AI SUMMARY REFERENCES EMBODIED MULTI-AGENT SYSTEM BACKGROUND AND MOTIVATION RECENT PROGRESS MULTI-AGENT FOR EMBODIED TASKS MULTI-AGENT STRATEGY LEARNING RELATED DATASETS ARCHITECTURES AND CHALLENGES PART II THEORY AND METHODS PERCEPTION-ACTION LOOP IN EMBODIED MULTI-AGENT SYSTEM CENTRALIZED PERCEPTION-ACTION LOOP IN HOMOGENEOUS EMBODIED MULTI-AGENT SYSTEM PROBLEM FORMULATION PERCEPTION-ACTION FRAMEWORK SIMULATION EXPERIMENTS DECENTRALIZED PERCEPTION-ACTION LOOP IN HOMOGENEOUS EMBODIED MULTI-AGENT SYSTEM PERCEPTION-ACTION LOOP IN HETEROGENEOUS EMBODIED MULTI-AGENT SYSTEM QUALITATIVE ANALYSIS ABLATION EXPERIMENTS EMBODIED COOPERATION IN MULTI-AGENT SYSTEM RELATED WORKS EMBODIED COOPERATION HRL IN LOCOMOTION CONTROL PROBLEM STATEMENT DISTRIBUTED HIERARCHICAL REINFORCEMENT LEARNING TASK DECOMPOSITION AND HIERARCHICAL LEARNING DISTRIBUTED SCALABLE HRL SPATIOTEMPORAL MEMORY RECURRENCE TRAINING CURRICULUM EXPERIMENTS ENVIRONMENTS CONSTRUCTION RESULTS OF DISTRIBUTED HRL ABLATION SCALABILITY EXPERIMENTS TYPICAL RESULT COMPARISON TO OTHER MARL ALGORITHMS LIMITATIONS COMPETITIVE LEARNING IN EMBODIED MULTI-AGENT SYSTEM LEARNING FROM COMPETITIVE INFORMATION COMPETITIVE FRAMEWORK FOR INTERACTION LEARNING EMPIRICAL RESULTS ADVERSARIAL SKILL LEARNING FOR ROBUST MANIPULATION BACKGROUND AND PRELIMINARY ADVERSARIAL SKILL LEARNING CONCLUSION COEVOLVE BODY AND BRAIN IN CONFRONTATION CONTINUOUS SELF-PRACTICE INTERACTIVE TRAINING MORPH EVOLUTION AND TACTICS TYPICAL EXAMPLES ILLUSTRATING THE ROLES OF MORPHOLOGICAL EVOLUTION LARGE LANGUAGE MODEL FOR EMBODIED MULTI-AGENT SYSTEM AD HOC TEAMWORK PROBLEM IN EMBODIED MULTI-AGENT SYSTEM AD HOC FRAMEWORK IN EMBODIED MULTI-AGENT SYSTEM PRELIMINARY INFORMATION COMMUNICATION VISUAL SEMANTIC PERCEPTION DYNAMIC SUB-TASK PLANNER WITH IROT DYNAMIC SUBSKILL PLANNER SUBSKILL EXECUTION STATE ASSESSMENT GENERALIZATION TO DIFFERENT TASKS DATASET HETEROGENEOUS TEAMMATES POLICY AD HOC AGENT POLICY IMPLEMENTATION DETAILS EVALUATION METRICS QUANTITATIVE PERFORMANCE QUALITATIVE PERFORMANCE PART III APPLICATIONS SIMULATION PLATFORM FOR EMBODIED COLLABORATION BETWEEN HUMAN AND ROBOTS RELATED WORK HUMAN-ROBOT COLLABORATION SIMULATORS AND BENCHMARKS FOR EMBODIED INTELLIGENCE SYSTEM FRAMEWORK SYSTEM OVERVIEW SUPPORTED APIS USER INTERFACE AND VR SUPPORTS BENCHMARKING HRC WITH EVERYDAY TASKS OBJECT-CENTRIC INTERACTIVE COMMUNICATION GENERATING EVERYDAY TASKS WITH SCENE PRIORS BASELINES EXPERIMENTAL SETUP RESULTS CASE STUDY SYSTEM EXTENSIONS MUTLI-ROBOT SETTING MORE COMPLEX TASKS MORE ROBOT ALGORITHMS APPLICATION OF EMBODIED MULTI-AGENT SYSTEM APPLICATION OF HOMOGENEOUS EMBODIED MULTI-AGENT FOR AUTONOMOUS TASK ASSIGNMENT DEPLOYMENT SETTING DEMONSTRATION APPLICATION OF HETEROGENEOUS EMBODIED MULTI-AGENT FOR ROOM TIDYING-UP CONTENTS PART IV CONCLUSIONS CONCLUSIONS AND FUTURE DIRECTIONS FUTURE DIRECTIONS ACTIVE COLLABORATION INTERACTIVE LEARNING EMBODIED HUMAN-AGENT COLLABORATION